System, method, and apparatus for image-based plaque analysis and risk assessment.

Abstract

This application relates to a system, method, and apparatus for image-based analysis of plaque. In some embodiments, the methods described herein can be used for developing treatment plans. Treatment plans may include local treatment, systemic treatment, or both. In some embodiments, the approaches described herein can be used for stent selection. In some embodiments, the approaches described herein can be used for surgical planning, including robot-assisted surgical planning. In some embodiments, the approaches described herein can be used for image normalization. In some embodiments, the approaches described herein can be used to identify a threshold for plaque calcification. In some embodiments, the approaches described herein can be used to identify thin-capsule fibrous atheroma. In some embodiments, the approaches described herein can be used for reconstruction of the coronary artery tree. Some embodiments are intended for risk stratification of coronary artery disease.

Description

Technical Field

[0001] [Priority and Related Applications] This application claims the benefit of priority of U.S. Provisional Application No. 63 / 557,405, filed Feb. 23, 2024; U.S. Provisional Application No. 63 / 557,401, filed Feb. 23, 2024; U.S. Provisional Application No. 63 / 606,584, filed Dec. 5, 2023; U.S. Provisional Application No. 63 / 597,528, filed Nov. 9, 2023; U.S. Provisional Application No. 63 / 582,792, filed Sep. 14, 2023; U.S. Provisional Application No. 63 / 519,220, filed Aug. 11, 2023; U.S. Provisional Application No. 63 / 557,396, filed Feb. 23, 2024; and U.S. Provisional Application No. 63 / 499,602, filed May 2, 2023. The content of each of the above applications is hereby incorporated by reference in its entirety.

[0002] [Technical Field] This application relates to non-invasive image-based plaque analysis and risk determination. Some embodiments relate to treatment planning for plaques. Some embodiments relate to guidance for local and / or systemic plaque treatment using image-based plaque analysis. Some embodiments relate to normalization of medical images for plaque analysis, vascular analysis, or both. Some embodiments relate to risk stratification of coronary artery disease. Some embodiments relate to reconstruction of coronary artery trees using one or more image series. Some embodiments relate to methods for identifying calcified plaques using coronary computed tomography angiography. Some embodiments relate to adjustment of variable plaque classification thresholds. Some embodiments relate to methods for detecting thin-cap fibroatheromas using coronary computed tomography angiography. Some embodiments relate to stent selection and surgical planning.

Summary of the Invention

[0003] The approaches described in this section are potential approaches, but not necessarily approaches that have been conceived or pursued previously. Therefore, unless otherwise indicated, it should not be assumed that any of the materials described in this section qualify as prior art simply because they are included in this section.

[0004] Coronary artery plaque is a serious health problem. However, current diagnostic and treatment approaches have problems. Therefore, improved approaches are needed.

[0005] The various embodiments described herein relate to systems, apparatus, and methods for image-based plaque analysis and risk assessment. In particular, in some embodiments, the systems, apparatus, and methods described herein relate to the analysis of one or more regions of plaque, such as coronary artery plaque, based on one or more measurements of distance, volume, density, radiation density, shape, form, degree of embedding, and / or axis.

[0006] The various embodiments described herein relate to systems, apparatus, and methods for using image-based plaque analysis to guide local and / or systemic plaque treatment. In particular, in some embodiments, the systems, devices, and methods described herein relate to image-based analysis of one or more regions of plaque, such as coronary artery plaque, based on one or more measurements of distance, volume, density, radiation density, shape, morphology, degree of embedding, and / or axis. For example, in some embodiments, the systems, devices, and methods described herein relate to using one or more such analyses of plaque to determine and / or guide local and / or systemic treatment of a patient's plaque. In some embodiments, the systems, apparatus, and methods described herein are configured to utilize machine learning (ML) and / or artificial intelligence (AI).

[0007] The various embodiments described herein relate to systems, apparatus, and methods for normalizing medical images for plaque and / or vascular analysis. Some embodiments relate, in particular, to image processing algorithms that can be used to compare images taken with different image acquisition parameters.

[0008] The various embodiments described herein relate to systems, apparatus, and methods for image-based coronary artery disease (CAD) risk stratification.

[0009] The various embodiments described herein relate to systems, apparatus, and methods for reconstructing coronary artery trees.

[0010] The various embodiments described herein relate to systems, apparatus, and methods for plaque classification, and more particularly to automatic adjustment of plaque classification thresholds in computed tomography images.

[0011] The various embodiments described herein relate to systems, apparatus, and methods for detecting fibrous hemangiomas using coronary computed tomography angiography.

[0012] The various embodiments described herein relate to systems, apparatus, and methods for analyzing intravascular lesions. Some embodiments relate to stent selection. Some embodiments relate to surgical planning. Some embodiments relate to robotic surgical planning.

[0013] For the purposes of this abstract, specific aspects, advantages, and novel features of the present invention are described herein. It should be understood that not all such advantages can necessarily be achieved according to any particular embodiment of the present invention. Therefore, for example, a person skilled in the art will recognize that the present invention can be embodied or practiced in a manner that achieves one advantage or group of advantages as taught herein without necessarily achieving other advantages that can be taught or suggested herein.

[0014] All of these embodiments are intended to be within the scope of the invention disclosed herein. These embodiments and other embodiments will be readily apparent to those skilled in the art from the following detailed description with reference to the accompanying drawings, but the invention is not limited to the specific embodiments (s) disclosed.

Brief Description of the Drawings

[0015] A detailed description of embodiments of the present invention will be described with reference to the accompanying drawings.

[0016] [Figure 1] It is a schematic diagram of an exemplary embodiment of a system including a processing system configured to characterize a coronary plaque.

[0017] [Figure 2] It is a schematic diagram showing an example of myocardium and its coronary artery.

[0018] [Figure 3] It is a diagram showing an example of a series of images generated from a scan along a coronary artery, including a selected image of a part of the coronary artery, and how the image data corresponds to Hounsfield scale values.

[0019] [Figure 4] It is a block diagram showing a computer system in which various embodiments can be implemented.

[0020] [Figure 5A] It is a block diagram showing an example of a process for identifying features of medical images using artificial intelligence or machine learning.

[0021] [Figure 5B] It is a schematic diagram showing an example of a neural network for determining patient characteristics based on medical images.

[0022] [Figure 5C]A flowchart for training an artificial intelligence or machine learning model according to some embodiments is shown.

[0023] [Figure 5D] Examples of training and using an AI / ML model according to some embodiments are shown.

[0024] [Figure 6] A block diagram showing an embodiment of a computer hardware system configured to execute software for implementing one or more embodiments of the systems, methods, and apparatuses described herein.

[0025] [Figure 7] A block diagram showing an exemplary process for distinguishing true plaque regression from pseudo-plaque regression according to some embodiments.

[0026] [Figure 8] A block diagram showing an exemplary process for determining a recommended treatment according to some embodiments.

[0027] [Figure 9] A flowchart showing an example of an image normalization process according to some embodiments.

[0028] [Figure 10] A flowchart showing an exemplary process for generating an image processing algorithm according to some embodiments.

[0029] [Figure 11] A block diagram showing an exemplary process for determining risk stratification according to some embodiments.

[0030] [Figure 12] A flowchart showing an exemplary process for reconstructing a coronary artery tree according to some embodiments.

[0031] [Figure 13] This is a diagram illustrating the reconstructed coronary artery tree and the series used to generate it.

[0032] [Figure 14] This flowchart shows an exemplary process for adjusting the plaque classification threshold in coronary computed tomography angiography images according to several embodiments.

[0033] [Figure 15] This diagram schematically illustrates the effect of the plaque calcification threshold in several embodiments.

[0034] [Figure 16] This diagram schematically illustrates various possible thresholds for plaque according to several embodiments.

[0035] [Figure 17] This table shows examples of plaque calcification thresholds in relation to lumbar contrast imaging and kilovolts.

[0036] [Figure 18] This flowchart shows an exemplary process for determining a threshold Hounsfield unit value to distinguish low-attenuation plaques from other non-calcifying plaques, according to several embodiments.

[0037] [Figure 19] A flowchart shows an exemplary process for determining a threshold Hounsfield unit value for identifying thin capsular fibrous hemangiomas according to several embodiments.

[0038] [Figure 20] This flowchart shows an exemplary process for identifying thin-capsule fibrous hemangiomas in coronary computed tomography angiography images using machine learning, according to several embodiments.

[0039] [Figure 21A] This section schematically shows examples of coronary computed tomography angiography images that do not show thin capsular fibromas and coronary computed tomography angiography images that show thin capsular fibromas. [Figure 21B] This section schematically shows examples of coronary computed tomography angiography images that do not show thin capsular fibromas and coronary computed tomography angiography images that show thin capsular fibromas.

[0040] [Figure 22A] This diagram schematically illustrates examples of coronary computed tomography angiography images with different cutoff thresholds for low-attenuation plaques. [Figure 22B] This diagram schematically illustrates examples of coronary computed tomography angiography images with different cutoff thresholds for low-attenuation plaques.

[0041] [Figure 23] Examples of the separation of thin-capsulated fibrous hemangiomas and thick-capsulated fibrous hemangiomas at different Hounsfield unit thresholds are shown in several embodiments.

[0042] [Figure 24] This is a diagram illustrating examples of user interfaces according to several embodiments.

[0043] [Figure 25] This is a diagram showing examples of different user interfaces using several embodiments.

[0044] [Figure 26] This flowchart shows an example process for generating surgical plans in several embodiments.

[0045] The technology described herein will become more apparent to those skilled in the art by examining the detailed description in conjunction with the drawings. Embodiments or representations illustrating aspects of the invention are illustrated as examples, and the same reference may indicate similar elements. Although the drawings depict various embodiments for illustrative purposes, those skilled in the art will recognize that alternative embodiments can be adopted without departing from the principles of the technology of the invention. Thus, although specific embodiments are shown in the drawings, the technology is subject to various modifications. [Modes for carrying out the invention]

[0046] While several embodiments, examples, and illustrations are disclosed below, it will be understood by those skilled in the art that the invention described herein extends beyond the specifically disclosed embodiments, examples, and illustrations, including other uses of the invention and their apparent modifications and equivalents. Embodiments of the invention are described with reference to the accompanying drawings, where similar figures refer to similar elements throughout. The terms used in the descriptions presented herein are not intended to be constrained or restrictive simply because they are used in conjunction with the detailed descriptions of specific embodiments of the invention. In addition, embodiments of the invention may possess several novel features, and no single feature is the sole cause of its desirable attributes or essential for carrying out the invention described herein.

[0047] "Plaque" or "area of ​​a plaque" or "one or more areas of a plaque" may be simply referred to as "plaque" for ease of reference unless explicitly or otherwise indicated by the context. For example, in some embodiments, the systems, apparatus, and methods described herein relate to plaque analysis based on one or more of the following: the distance between the plaque (e.g., coronary artery plaque) and the vessel wall; the distance between the plaque and the luminal wall; the length of the plaque along the anterior-posterior axis; the length of the plaque along the transverse axis; the area or volume of low-density uncalcified plaque; the area or volume of uncalcified plaque; the area or volume of calcified plaque; the total plaque area or volume; the ratio of one or more of the area or volume of low-density uncalcified plaque; the area or volume of uncalcified plaque; the area or volume of calcified plaque; or the degree of embedding of any of the plaques, including the total plaque area or volume, low-density uncalcified plaque, uncalcified plaque, calcified plaque, total plaque, and / or such. In some embodiments, the systems, apparatus, and methods described herein are configured to determine the risk of coronary artery disease (CAD) and / or major adverse cardiovascular events (MACE), such as myocardial infarction (MI), based on one or more plaque analyses described herein. In some embodiments, the systems, apparatus, and methods described herein are configured to generate proposed treatments and / or graphical representations based on the determined CAD risk and / or one or more plaque analyses described herein. In some embodiments, the systems, apparatus, and methods described herein may be configured to generate multiple treatment options from which an intervener can select.

[0048] In some embodiments, the systems, methods, and apparatus described herein are configured to analyze one or more coronary computed tomography (CCTA) images to identify one or more high-risk plaques or atherosclerosis. In some embodiments, high-risk plaques or atherosclerosis may be identified if one or more high-risk factors are present, including, for example, high volume, burden, composition, density (also referred herein to as material density), radiation density, and / or such. In some embodiments, high-risk plaques or atherosclerosis may be identified at the patient level, the lesion level, or any other intermediate level.

[0049] In some embodiments, the systems, methods, and apparatus described herein may be configured to analyze the total plaque volume of a patient and / or the presence and / or prevalence or extent of high-risk plaques. For example, in some embodiments, high-risk plaques may be identified based on low attenuation, low material density, low radiation density, and / or high lesion-level plaque volume. In some embodiments, the systems, methods, and apparatus described herein may be configured to determine or generate a lesion-level risk score. In some embodiments, the lesion-level risk score may be used to identify one or more focal lesions that have a poor prognosis and / or a high or relatively high risk of becoming the causative lesion at a future MI or other MACE.

[0050] [introduction] Coronary heart disease affects more than 17.6 million Americans. There are generally two current trends in the treatment of cardiovascular health problems. First, physicians generally consider a patient's cardiovascular health from a macro level. For example, they analyze the patient's biochemistry, blood components, and / or biomarkers to determine if there are high levels of cholesterol components in the patient's bloodstream. In response to high levels of cholesterol, some physicians prescribe one or more medications, such as statins, as part of a treatment plan to reduce what is perceived as high levels of cholesterol components in the patient's bloodstream.

[0051] Currently, a second common trend in treating cardiovascular health problems is for physicians to use angiography to assess a patient's cardiovascular health and identify major blockages in various arteries. If major blockages are found in various arteries, physicians may perform angioplasty, guiding a balloon catheter to the narrowed point of the vessel. After properly positioning the balloon, it is inflated to press or flatten plaque and fat into the arterial wall and / or stretch the artery, increasing blood flow within the vessel and / or to the heart. In some cases, the balloon is used to place and expand a stent within the vessel to compress plaque and / or maintain the opening of the vessel, allowing more blood to flow. Approximately 500,000 cardiac stent procedures are performed annually in the United States.

[0052] However, a recent $100 million federally funded study questions whether current trends in cardiovascular disease treatment are the most effective for all types of patients. The recent study, involving more than 5,000 patients with moderate to severe stable heart disease from 320 sites in 37 countries, presented new evidence suggesting that stents and bypass surgery may be less effective than medication combined with lifestyle modifications for patients with stable heart disease. Therefore, it may be more beneficial for patients with stable heart disease to avoid invasive surgical procedures such as angioplasty and bypass surgery, and instead be prescribed heart medications like statins, along with lifestyle modifications such as regular exercise. This new approach could impact thousands of patients worldwide. In the United States, an estimated 500,000 cardiac stent procedures are performed annually, with one-fifth estimated to be for patients with stable heart disease. Furthermore, it is estimated that 25% of the estimated 100,000 people with stable heart disease—approximately 25,000—do not experience chest pain. Therefore, more than 20,000 patients each year could potentially avoid invasive surgery and its resulting complications.

[0053] A more complete understanding of a patient's cardiovascular disease can be crucial in determining whether they should forgo invasive surgery and opt for drug therapy instead, and / or in developing a more effective treatment plan. Specifically, a better understanding of the patient's arterial vascular health is beneficial. For example, it is helpful to know whether a patient's plaque accumulation is mostly fat accumulation or mostly calcification accumulation. In the former case, treatment with cardiac medications such as statins may be justified, while in the latter case, the patient should receive further regular monitoring without being prescribed cardiac medications or having a stent implanted. However, if the plaque accumulation is severe enough to cause severe stenosis or ductus arteriosus stenosis that blocks blood flow to the myocardium, invasive angioplasty with stent placement may be necessary. Sudden cardiac death (SCD) is one of the leading causes of natural death in the United States, accounting for approximately 325,000 adult deaths annually and nearly half of all deaths from cardiovascular disease. SCD is twice as common in men as in women. Generally, SCD develops between the mid-30s and mid-40s. In more than 50% of cases, sudden cardiac arrest occurs without warning.

[0054] For millions of people suffering from heart disease, it is necessary to better understand the overall health of the arterial vessels within patients, not only by knowing the blood chemistry or content of the blood flowing through such arteries. For example, in some embodiments of the systems, apparatus, and methods disclosed herein, arteries with “good” or stable plaque or plaque consisting of hardened calcified contents are considered not life-threatening to patients, while arteries with “bad” or unstable plaque or plaque consisting of fatty substances are considered more life-threatening because such bad plaque may rupture within the artery, releasing such fatty substances into the artery. The release of such fatty substances into the bloodstream can cause inflammation and lead to thrombosis. If a thrombus forms in an artery, blood cannot be delivered to the heart muscle, which can lead to a heart attack or other heart disease. Furthermore, in some cases, it is generally more difficult for blood to flow through an accumulation of fatty plaque than through an accumulation of calcified plaque. Therefore, it is necessary to better understand and analyze the arterial walls of patients.

[0055] Furthermore, while blood tests and drug regimens are useful in mitigating cardiovascular health problems and alleviating cardiovascular events (e.g., heart attacks), such treatments are not perfect in that they may misidentify and / or fail to pinpoint significant cardiovascular risk areas. For example, simply analyzing a patient's blood chemistry is unlikely to identify that the patient has arterial vessels with a large accumulation of fatty deposits (bad plaque) along the vessel walls. Similarly, while angiography is useful in identifying the sites of stenosis and narrowing of vessels, it may not be able to clearly identify areas where bad plaque is significantly accumulating in the arterial vessel walls. Such areas of bad plaque accumulation in the arterial vessel walls can be indicators of a patient at high risk of developing cardiovascular events such as heart attacks. In certain circumstances, areas with bad plaque may lead to rupture, releasing fatty substances into the arterial bloodstream, potentially resulting in the formation of a blood clot in the artery. When a blood clot forms in an artery, it can cut off blood flow to the heart tissue, potentially causing a heart attack. Therefore, there is a need for new techniques to analyze arterial walls and / or identify areas within arterial walls that have plaque accumulation, whether malignant or not.

[0056] In some embodiments, the systems, apparatus, and methods described herein are configured to utilize non-invasive medical imaging techniques, such as CT imaging or CCTA, and can be input into a computer system configured to automatically and / or dynamically analyze medical images to identify one or more coronary arteries and / or plaques therein. For example, in some embodiments, the system may be configured to automatically and / or dynamically analyze medical images to identify, quantify, and / or classify one or more coronary arteries and / or plaques using one or more machine learning and / or artificial intelligence algorithms. In some embodiments, the system may be further configured to use the identified, quantified, and / or classified one or more coronary arteries and / or plaques to generate treatment plans, track disease progression, and / or generate patient-specific medical reports, for example, using one or more artificial intelligence and / or machine learning algorithms. In some embodiments, the system may be further configured to dynamically and / or automatically generate visualizations of the identified, quantified, and / or classified one or more coronary arteries and / or plaques, for example, in the form of a graphical user interface. Furthermore, in some embodiments, the system may be configured to utilize a normalization device comprising one or more sections of one or more materials in order to calibrate medical images obtained from different medical imaging scanners and / or different scan parameters or environments.

[0057] As will be described in more detail, the systems, apparatus, and methods described herein enable automated and / or dynamic quantitative analysis of various parameters relating to plaque, cardiovascular arteries, and / or other structures. More specifically, in some embodiments described herein, medical images of a patient, such as coronary CT images or CCTA, can be acquired in a medical facility. Rather than a physician performing a visual inspection or general assessment of the patient, the medical images are sent to a backend main server, in some embodiments configured to perform one or more analyses of them in a reproducible manner. Thus, in some embodiments, the systems, methods, and apparatus described herein can provide quantified measurements of one or more features of a coronary CT image using automated and / or dynamic processes. For example, in some embodiments, the main server system can be configured to identify one or more vessels, plaque, fat, and / or one or more measurements thereof from the medical image. Based on the identified features, in some embodiments, the system may be configured to generate one or more quantified measurements from the raw medical image, such as the radiation density of one or more regions of plaque, identification of stable and / or unstable plaque, its volume, its surface area, geometric shape, its heterogeneity, and / or such. In some embodiments, the system can also generate one or more quantified measurements of blood vessels from raw medical images, such as diameter, volume, morphology, and / or similar. Based on the identified features and / or quantified measurements, in some embodiments, the system may be configured to use raw medical images to generate risk and / or disease status assessments of plaque-based diseases or conditions, such as atherosclerosis, stenosis, and / or ischemia, and / or track their progression. Furthermore, in some embodiments, the system may be configured to generate GUI visualizations of one or more identified features and / or quantified measurements, such as quantized color mappings of different features.In some embodiments, the systems, apparatus, and methods described herein are configured to utilize medical image-based processing to assess the risk of cardiovascular events, major adverse cardiovascular events (MACE), rapid plaque progression, and / or non-response to medication for a subject. In particular, in some embodiments, the system may be configured to automatically and / or dynamically assess such health risks of a subject by analyzing only non-invasively obtained medical images. In some embodiments, one or more of the processes can be automated using artificial intelligence (AI) and / or machine learning (ML) algorithms. In some embodiments, one or more of the processes described herein can be performed within minutes in a reproducible manner. This is in contrast to existing means today that do not produce reproducible prognoses or assessments, require enormous amounts of time, and / or invasive procedures. In some embodiments, the systems, methods, and apparatus described herein are configured to include and / or utilize one or more of such techniques described in U.S. Patent Application Publication No. US 2021 / 0319558, which is incorporated herein in whole by reference.

[0058] Thus, in some embodiments, the systems, apparatus, and methods described herein can provide physicians and / or patients with specific quantifiable and / or measurable data related to patient plaque and / or ischemia that are not currently present. In some embodiments, such a detailed level of plaque parameters quantified from image processing and downstream analysis results can provide a more accurate and useful tool for assessing patient health and / or risk in entirely novel ways.

[0059] A method for identifying high-risk plaques is disclosed, utilizing the volumetric features of coronary artery plaque and perivascular adipose tissue data obtained by computed tomography (CT). Volumetric characterization of coronary artery plaque and perivascular adipose tissue enables the determination of the inflammatory state of plaques by CT scans. This is useful for the diagnosis, prognosis, and treatment of coronary artery disease. Certain exemplary embodiments are illustrated in the drawings and described in detail herein, but these embodiments are subject to various modifications and alternative forms. The exemplary embodiments are not intended to limit themselves to the specific forms disclosed, but rather to cover all modifications, equivalents, and alternatives that fall within the scope of the exemplary embodiments.

[0060] In this specification, terms such as "first," "second," etc., may be used to describe various elements, but it will be understood that these elements should not be limited by these terms. These terms are used merely to distinguish one element from another. For example, without departing from the scope of the exemplary embodiments, a first element may be referred to as a second element, and similarly, a second element may be referred to as a first element. As used herein, the terms "and / or" include all combinations of one or more of the related enumerated items.

[0061] The terms used herein are for illustrative purposes only and are not intended to limit the exemplary embodiments. Where used herein, the singular forms “a,” “an,” and “the” are intended to include the plural form unless the context clearly indicates otherwise. Where used herein, the terms “comprises,” “comprising,” “includes,” and / or “including” identify the presence of a described feature, integer, step, operation, element, and / or component, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. Where herein, the term “and / or” identifies not only individual items but also all combinations thereof.

[0062] Unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by those skilled in the art in the field to which the exemplary embodiments belong. Furthermore, terms as defined in commonly used dictionaries should be interpreted as having the meaning consistent with their meaning in the context of the relevant art, and should not be interpreted in an idealized or overly formal sense unless expressly defined herein. It should also be noted that in some alternative embodiments, the illustrated functions / actions may occur in a different order than that shown. For example, two functions / actions shown consecutively may actually be performed simultaneously, or, depending on the functions / actions involved, may be performed in reverse order.

[0063] In the drawings, the dimensions of layers and areas are exaggerated for clarity. Furthermore, when a layer (or structure) is described as "on top of" another layer or structure, it should be understood that it may be directly on top of the other layer or substrate, or there may be an intervening layer. Similarly, when a layer is described as "below" another layer, it may be directly beneath it, or there may be one or more intervening layers. In addition, when a layer is described as "between" two layers, it should be understood that it may be the only layer between the two layers, or there may be one or more intervening layers. Similar reference figures refer to similar elements throughout.

[0064] In some embodiments, the system may be configured to characterize a particular area of ​​plaque as high-risk or low-density non-calcified plaque when the radiation intensity of the image pixel or voxel corresponding to that area of ​​plaque is between about -189 and about 30 Houndsfield units (HU). In some embodiments, the system may be configured to characterize a particular area of ​​plaque as non-calcified plaque when the radiation intensity of the image pixel or voxel corresponding to that area of ​​plaque is between about 31 and about 350 HU. In some embodiments, the system may be configured to characterize a particular area of ​​plaque as calcified plaque when the radiation intensity of the image pixel or voxel corresponding to that area of ​​plaque is between about 351 and about 2500 HU.In some embodiments, the lower and / or upper limits of the Hounsfield unit boundary threshold for determining whether a plaque corresponds to one or more of low-density non-calcified plaques, non-calcified plaques, and / or calcified plaques are approximately -1000 HU, approximately -900 HU, approximately -800 HU, approximately -700 HU, approximately -600 HU, approximately -500 HU, approximately -400 HU, approximately -300 HU, approximately -200 HU, approximately -190 HU, approximately -180 HU, approximately -170 HU, approximately -160 HU, approximately -150 HU, approximately -140 HU, approximately -130 HU, approximately -120 HU, -110HU, -100HU, -90HU, -80HU, -70HU, -60HU, -50HU, -40HU, -30HU, -20HU, -10HU, 0HU, 10HU, 20HU, 30HU, 40HU, 50HU, 60HU, About 70HU, About 80HU, About 90HU, About 100HU, About 110HU, About 120HU, About 130HU, About 140HU, About 150HU, About 160HU, About 170HU, About 180HU, About 190HU, About 200HU, About 210HU, About 220HU, About 230HU, 240HU, 250HU, 260HU, 270HU, 280HU, 290HU, 300HU, 310HU, 320HU, 330HU, 340HU, 350HU, 360HU, 370HU, 380HU, 390HU, 400HU, Approximately 410HU, approximately 420HU, approximately 430HU, approximately 440HU, approximately 450HU, approximately 460HU, approximately 470HU, approximately 480HU, approximately 490HU, approximately 500HU, approximately 510HU, approximately 520HU, approximately 530HU, approximately 540HU, approximately 550HU, approximately 560HU, approximately 570HU, approximately 580HU, Approximately 590 HU, approximately 600 HU, approximately 700 HU, approximately 800 HU, approximately 900 HU, approximately 1000 HU, approximately 1100 HU, approximately 1200 HU, approximately 1300 HU, approximately 1400 HU, approximately 1500 HU, approximately 1600 HU, approximately 1700 HU, approximately 1800 HU, approximately 1900 HU, approximately 2000 HU, approximately 2100 HU, approximately 2200 HU, approximately 2300 HU, approximately 2400 HU, approximately 2500 HU, approximately 2600 HU, approximately 2700 HU, approximately 2800 HU, approximately 2900 HU, approximately 3000 HU, approximately 3100 HU, approximately 3200 HU, approximately 3300 HU, approximately 3400 HU, approximately 3500 HU, and / or approximately 4000 HU, possibly.

[0065] [Outline of an example processing system for evaluating coronary artery plaque] This disclosure includes methods and systems for using data generated from images collected by scanning a patient's arteries to identify coronary artery plaques that are at high risk of causing future heart attacks or acute coronary syndromes. In particular, the properties of perivascular coronary adipose tissue, coronary artery plaques, and / or coronary artery lumen, as well as the relationship between these properties, are discussed to determine a method for identifying coronary artery plaques that are likely to be associated with future ACS, heart attacks, and death. The images used to generate the image data may be CT images, CCTA images, or images generated using any applicable technique capable of depicting the relative density of coronary artery plaques, perivascular adipose tissue, and coronary artery lumen. For example, CCTA images may be used to generate two-dimensional (2D) or volumetric (three-dimensional (3D)) image data, which may be analyzed to determine specific properties related to the radiation density of coronary artery plaques, perivascular adipose tissue, and / or coronary artery lumen. In some embodiments, the Hounsfield scale is used to provide a measure of the radiation density of these properties. As is well known, the Haunsfield unit represents any unit of X-ray attenuation used in CT scans. Each pixel (2D) or voxel (3D) of a feature in image data can be assigned an irradiance value on the Haunsfield scale, and these values ​​characterizing the feature can be analyzed.

[0066] In various embodiments, the processing of image information may include the following: (1) Determine scan parameters (e.g., mA (milliamperes), kVp (peak kilovolts)); (2) Determine scan image quality (e.g., noise, signal-to-noise ratio, contrast-to-noise ratio); (3) Measure scan-specific coronary lumen density (e.g., from the distal point of the coronary artery wall to the proximal point of the coronary artery wall to the distal point of the coronary artery, and from the central to the lateral position of the coronary artery (e.g., lateral relative to radial distance from the coronary artery)); (4) Measure scan-specific plaque density (e.g., abruptness of changes from high to low or low to high within the plaque, from the center to the lateral); and (5) Measure scan-specific perivascular coronary fat density (from near to far from the artery) as a function of three-dimensional shape.

[0067] From these measurements, which are unknowable to the generally known characteristics of atherosclerosis that cause ischemia, the systems and methods of some embodiments described herein can be determined to have several features, including but not limited to one or more of the following: 1. The ratio of luminal attenuation to plaque attenuation, a volume model of the scan-specific attenuation density gradient within the lumen, adjusts for the decrease in luminal density across the entire plaque lesion, which is functionally important in terms of risk value; 2. The ratio of plaque attenuation to fat attenuation; among the subset of plaques considered "calcified," plaques with high radiation density are considered lower risk; 3. The ratio of lumen attenuation / plaque attenuation / fat attenuation; 4. The ratio of #1-3 as a function of the 3D shape of arteriosclerosis, which may include 3D texture analysis of plaque; 5. Show the three-dimensional volume shape and path of the lumen, along with the decay density from the start to the end of the lumen; 6. Analyze the plaques before and after a given plaque, as well as the type of plaque, to further understand the risk. 7. To obtain a better absolute indicator of high-risk plaques (low-concentration plaques), "high-risk plaques" are determined by "subtracting" calcified (high-concentration) plaques. In other words, in this particular embodiment, calcified plaques are identified and excluded from further analysis of the plaque for the purpose of identifying high-risk plaques. Other characteristics can also be identified.

[0068] By analyzing the above characteristics / indicators together with other indicators, the risk of plaque being associated with future heart attacks, ACS, ischemia, or death can be assessed. This can be done by developing and / or validating traditional risk scores, or by machine learning methods. Factors for analysis from metrics likely to be associated with heart attacks, ACS, ischemia, or death may include one or more of the following: (1) the ratio of [bright lumen:dark plaque], (2) the ratio of [dark plaque:bright fat], (3) the ratio of [bright lumen:dark plaque:bright fat], or (4) the low ratio of [dark lumen:dark myocardium in one vascular region] / [lumen:myocardium in another vascular region]. Some improvements in the disclosed methods and systems may include one or more of the following: (1) using numerical values ​​from the ratios of [lumen:plaque], [plaque:fat] and [lumen:plaque:fat] instead of using qualitative definitions of atherosclerotic features; (2) using the [lumen:plaque attenuation] ratio to characterize scan-specific [lumen:plaque]; (3) using the scan-specific [plaque:fat attenuation] ratio to characterize plaques; (4) using the [lumen:plaque:fat circumference] ratio to characterize plaques; or (5) integrating pre- and post-plaque volume and type as the contribution of individual plaques to risk.

[0069] The characteristics of atherosclerotic plaques can change over time with medical treatment (e.g., colchicine or statins), and while some of these drugs may slow plaque progression, they also play a crucial role in accelerating plaque changes. Statins may have reduced overall plaque progression, but may have actually increased the progression of calcified plaques and decreased non-calcified plaques. This change is associated with a reduction in heart attacks or ACS or mortality, and the disclosed methods can be used to monitor the effect of medical treatment on plaque risk over time. Furthermore, these methods can be used to identify individuals whose atherosclerotic plaque characteristics or [lumen:plaque] / [plaque:fat] / [lumen:plaque:fat] ratio indicate a higher risk of rapid disease progression or malignant transformation. In addition, these methods can be applied to single plaques, or the entire cardiac atherosclerosis tracking can be used to monitor the risk of patients experiencing heart attacks (rather than attempting to identify specific plaques as the cause of future heart attacks). Tracking can be performed through an automated, collaborative registration process of patient-related image data over a specified period.

[0070] Figure 1 shows a schematic diagram of an example embodiment of system 100, which includes a processing system 120 configured to characterize coronary artery plaque. The processing system 120 may include one or more servers (or computers) 105, each configured with one or more processors. In some embodiments, the processing system 120 includes a non-temporary computer memory component for storing data and a non-temporary computer memory component for storing instructions executed by one or more processor data communication interfaces, the instructions configuring one or more processors to perform a method of analyzing image information. A more detailed example of a server / computer 105 is described with reference to Figure 6.

[0071] In some embodiments, system 100 also includes a network. Processing system 120 can communicate with network 125. Network 125 may include, as at least a part of network 125, the internet, a wide area network (WAN), a wireless network, etc. In some embodiments, processing system 120 is part of a “cloud” implementation and can be located anywhere that can communicate with network 125. In some embodiments, processing system 120 is located in geographically close proximity to the imaging facility that acquires and stores patient image data. In some embodiments, processing system 120 is located remotely from the location where patient image data is generated or stored.

[0072] Figure 1 also shows various computer systems and devices 130 (e.g., of an imaging facility) in system 100 that are involved in generating patient image data and may also be connected to network 125. One or more of the devices 130 may be located in an imaging facility, medical facility (e.g., a hospital, clinic, etc.) that generates images of the patient's arteries, or they may be personal computing devices of the patient or caregiver. For example, as shown in Figure 1, an imaging facility server (or computer) 130A may be connected to network 125. In this example, a scanner 130B in the imaging facility may also be connected to network 125. One or more other computer devices may also be connected to network 125. For example, a laptop 130C, a personal computer 130D, and / or an image information storage system 130E may also be connected to network 125 and can communicate with the processing system 120 and with each other via network 125.

[0073] In some embodiments, scanner 130B can be a computed tomography (CT) scanner that uses a rotating row of X-ray tubes and detectors to measure the attenuation of X-rays by different tissues in the body and form a corresponding image. In another embodiment, scanner 130B can use a rotating tube ("spiral CT") in which the entire X-ray tube and detectors rotate around a central axis of the area being scanned. In yet another embodiment, scanner 130B can utilize electron beam tomography (EBT). In yet another embodiment, scanner 130B can be a dual-source CT scanner having two X-ray tube systems. In yet another embodiment, scanner 130B can be a multi-source CT scanner having two or more X-ray tube systems. In yet another embodiment, scanner 130B can include a high-speed switching X-ray tube system. The methods and systems described herein can also use images from other CT scanners. In some embodiments, scanner 130B is a photon-counting CT scanner, a spectral CT scanner, or a dual-energy CT scanner. Photon-counting CT scanners, spectral CT scanners, multispectral CT scanners, or dual-energy CT scanners help provide more detailed, high-resolution images that better show small vessels, plaques, and other vascular pathologies, and allow for the determination of absolute material density rather than relative density. Generally, photon-counting detector CT scanners use an X-ray detector to count photons, quantify their energy, and determine the count of photons in several discrete energy bins. As a result, the contrast-to-noise ratio is higher and spatial resolution and spectral imaging are improved compared to conventional CT scanners. Each registered photon can be assigned to a specific bin according to its energy, and each pixel measures a histogram of the incident X-ray spectrum. This spectral information offers several advantages. Firstly, it can be used to quantitatively determine the material composition of each pixel in the reconstructed CT image, in contrast to the estimated mean attenuation coefficient obtained in conventional CT scans. Spectral / energy information can be used to remove beam hardening artifacts that occur with high linear attenuation in many materials, which shift the average energy of the X-ray spectrum to the higher energy side. Furthermore, using two or more energy bins allows for the identification of objects (bone, calcification, contrast agents, tissue, etc.). In some embodiments, images generated using a photon-counting CT scanner allow for plaque evaluation at different multicolor spectra (e.g., 100kVp, 120kVp, 140kVp, etc.) as well as different monochromatic energies, thereby altering the definition of non-calcified and calcified plaques compared to conventional CT scanners. Spectral CT scanners can generate CT scans using different X-ray wavelengths (or energies). Dual-energy CT scanners can use separate X-ray energies to detect two different energy ranges.In one example, a dual-energy CT scanner (also called spectral CT) can use X-ray detectors with separate layers to detect two different energy ranges ("dual layer"). In another example, a dual-energy CT scanner can use a single scanner to scan twice using two different energy levels (e.g., electron kVp switching). Images detected at each different energy level can be combined to form an image, or the images can be used separately to assess the patient's condition. Photon-counting CT scanners not only provide absolute material density but can also assess images that are "monochromatic," in contrast to typical CT which is a multicolor spectrum of light. As mentioned above, features depicted in images formed using a photon-counting CT scanner, spectral CT scanner, or dual-energy CT scanner (e.g., low-density non-calcified plaque, calcified plaque, non-calcified plaque) may have different radioactivity concentrations than those depicted in images formed from conventional CT scanners, meaning that such images may affect or alter the definitions of calcified and non-calcified plaques. However, the radiation density of calcified and non-calcified plaques, or other characteristics, depicted in images formed from photon-counting CT scanners, spectral CT scanners, or dual-energy CT scanners can be normalized to correspond to the densities of conventional CT scanners and to the densities disclosed herein. Thus, the radiation densities disclosed herein can be directly correlated to the radiation density of images produced by photon-counting CT scanners, spectral CT scanners, or dual-energy CT scanners, so that the systems and methods, analyses, plaque densities, etc., disclosed herein are directly applicable to images formed from photon-counting CT scanners, spectral CT scanners, or dual-energy CT scanners, and can be directly applied to images formed from photon-counting CT scanners, spectral CT scanners, or dual-energy CT scanners normalized to the radiation density of equivalent conventional CT scanners.

[0074] Information communicated from the device 130 to the processing system 120 via the network 125 may include image information 135. In various embodiments, the image information 135 may include 2D or 3D image data of a patient, scan information related to the image data, patient information, and other image or image-related information related to the patient. For example, the image information may include patient information including (one or more) characteristics of the patient, e.g., age, sex, body mass index (BMI), medication, blood pressure, heart rate, height, weight, race, whether the patient is a smoker or non-smoker, body type (e.g., "body type" or "physique"), medical history, diabetes, hypertension, history of coronary artery disease (CAD), dietary habits, medication history, family history of disease, information related to other image information previously collected, exercise habits, drinking habits, lifestyle information, test results, etc. In some embodiments, the image information may include patient identification information, e.g., patient's name, patient's address, driver's license number, social security number, or other indicators of patient identification. When the processing system 120 analyzes the image information 135, information related to the patient 140 may be communicated from the processing system 120 to the device 130 via the network 125. The patient information 140 may include, for example, a patient report. It may also include various patient information available from a patient portal that can be accessed by any of the devices 130.

[0075] In some embodiments, image information comprising multiple images of a patient's coronary arteries and patient information / characteristics may be provided from one or more devices 130 to one or more servers 105 of a processing system 120 via a network 125. In some embodiments, the processing system 120 is configured to generate coronary artery information using multiple images of the patient's coronary arteries to generate a two-dimensional and / or three-dimensional data representation of the patient's coronary arteries. In some embodiments, the processing system 120 analyzes the data representation to generate a patient report documenting the patient's health status and risks associated with coronary plaque. The patient report may include images and graphs of the patient's arteries in terms of the type of coronary plaque within or near the coronary arteries. Using machine learning techniques or other artificial intelligence techniques, the data representation of a patient's coronary arteries can be compared to data representations of other patients (e.g., stored in a database) to determine additional information about the patient's health. For example, based on the status of specific plaques in the patient's coronary arteries, it can be determined whether the patient is likely to suffer a heart attack or other coronary artery adverse effects. Also, additional information about the patient's CAD risk can be determined, for example.

[0076] Figure 2 is a schematic diagram showing an example of myocardium 225 and its coronary arteries. The coronary vascular system includes a complex network of vessels ranging from large arteries to arterioles, capillaries, veins, and so on. Figure 2 depicts a model of a portion of the coronary artery system that circulates blood to and within the heart, and includes the aorta 240 which supplies blood to several coronary arteries, such as the left anterior descending (LAD) artery 215, the left circumflex (LCX) artery 220, and the right coronary artery (RCA) artery 205, which will be discussed later. The coronary arteries supply blood to myocardium 225. Like all other tissues in the body, myocardium 225 requires oxygen-rich blood to function. It must also carry away oxygen-deficient blood. The coronary arteries wrap around the outside of myocardium 225. Smaller branches penetrate into myocardium 225 and carry blood. Examples of methods and systems described herein can be used to determine information relating to the blood flowing through the coronary arteries in any vessels extending therefrom. In particular, the described methods and system examples can be used to determine various pieces of information related to one or more portions of a coronary artery where plaque has formed, which can be used to determine the risks associated with such plaque, for example, whether plaque formation poses a risk of causing adverse events in the patient.

[0077] The right side 230 of the heart 225 is depicted on the left side (relative to the page) of Figure 2, and the left side 235 of the heart is depicted on the right side of Figure 2. The coronary arteries include the right coronary artery (RCA) 205, which extends downward from the aorta 240 along the right side 230 of the heart 225, and the left coronary artery (LMCA) 210, which extends downward from the aorta 240 on the left side 235 of the heart 225. The RCA 205 supplies blood to the right ventricle, right atrium, and the SA (sinoatrial) and AV (atrioventricular) nodes, which regulate the heart's rhythm. The RCA 205 branches into smaller branches such as the right posterior descending artery and acute marginal artery. Together with the left anterior descending artery 215, the RCA 205 helps supply blood to the middle or septum of the heart.

[0078] The left coronary artery (LMCA) 210 branches into two arteries: the left anterior interventricular branch (also known as the left anterior descending coronary artery (LAD) 215) and the left circumflex coronary artery 220. The LAD 215 supplies blood to the left anterior part of the heart. Occlusion of the LAD 215 is often called widow's infarction. The circumflex branch of the left coronary artery 220 surrounds the myocardium. Following the left portion of the coronary groove, the circumflex branch of the left coronary artery 220 supplies blood to the lateral and posterior parts of the heart, running first to the left, then to the right, and reaching almost to the posterior longitudinal groove.

[0079] Figure 3 shows an example of a series of images generated from a scan along the coronary arteries, including selected images of a portion of the coronary arteries, and how the image data may correspond to values ​​on the Hounsfield scale. As explained with reference to Figure 1, in addition to obtaining image data, scanning information including metrics related to the image data, and patient information including patient characteristics can also be collected.

[0080] A portion of the heart 225, the LMCA 210, and the LAD artery 215 are shown in the example in Figure 3. Along the LMCA 210 and a portion of the LAD artery 215, in this example, a set of images 305 can be collected from a first point 301 on the LMCA 210 to a second point 302 on the LAD artery 215. In some examples, image data can be obtained using non-invasive imaging techniques. For example, CCTA image data can be generated using a scanner for creating images of the heart in the coronary arteries and other vessels extending therefrom. The collected CCTA image data can then be used to generate a three-dimensional image model of the features contained in the CCTA image data (e.g., the right coronary artery 205, the left main coronary artery 210, the left anterior descending artery 215, the circumflex branch of the left coronary artery 220, the aorta 240, and other heart-related vessels appearing in the image data).

[0081] In various embodiments, different imaging methods can be used to acquire image data. For example, ultrasound or magnetic resonance imaging (MRI) may be used. In some embodiments, the imaging method includes the use of a contrast agent to help identify the structure of the coronary arteries, and the contrast agent is injected into the patient before the imaging procedure. The various imaging methods may each have advantages and disadvantages in use, including resolution and suitability for imaging coronary arteries. Imaging methods that can be used to acquire image data of coronary arteries are constantly being improved as improvements are made to the hardware (e.g., sensors and emitters) and software. The disclosed systems and methods intend to use CCTA image data and / or any other type of image data that can provide or transform representative 3D depictions of the coronary arteries, plaque contained within the coronary arteries, and perivascular fat located adjacent to the coronary arteries containing the plaque, so that attenuation values ​​or radiation density values ​​of the coronary arteries, plaque, and / or perivascular fat are obtained. In some embodiments, the imaging mode may include one or more of the following: CT, dual-energy computed tomography (DECT), spectral CT, photon counting CT, X-ray, ultrasound, echocardiography, intravascular ultrasound (IVUS), magnetic resonance imaging (MR), optical coherence tomography (OCT), nuclear medicine imaging, positron emission tomography (PET), single-photon emission tomography (SPECT), near-field infrared spectroscopy (NIRS), contrast-enhanced CT, and non-contrast CT.

[0082] Referring further to Figure 3, a specific image 310 of image data 305 is shown, which represents an image of a portion of the left anterior descending artery 215. Image 310 contains image information, and the smallest point of information manipulated by a system commonly referred to herein as a pixel is, for example, pixel 315 of image 310. The resolution of the imaging system used to acquire the image data affects the size of the smallest distinguishable feature in the image. Furthermore, subsequent image manipulation may also affect the size of the pixels. As an example, a digital image 310 may contain 4000 pixels in each row and 3000 pixels in each column. Pixel 315, and each pixel in image data 310 and image data 305, can be associated with a radiation density value corresponding to the density of pixels in the image. Illustratively shown in Figure 3 is the mapping of pixel 315 to a point on the Hounsfield scale 320, which is a quantitative scale for describing radiation density. The Haunsfield unit scale linearly transforms the original linear attenuation coefficient measurements to a scale where the radiant density of distilled water at standard pressure and temperature is defined as zero Haunsfield units (HU), and the radiant density of air at standard pressure and temperature is defined as -1000 HU. Figure 3 shows an example of mapping pixel 315 of image 310 to a point on the Haunsfield scale 320, but such associations between pixels and radiant density values ​​can also be performed with 3D data, for example, after generating a three-dimensional representation of the coronary arteries using image data 305.

[0083] Once data is acquired and rendered into a three-dimensional representation, various processes can be performed on the data to identify areas of analysis. For example, a three-dimensional depiction of coronary arteries may be segmented to define multiple parts of the arteries and thus identified within the data. In some embodiments, the data may be filtered (e.g., smoothed) in various ways to remove anomalies that are the result of scanning or various other errors. Various known methods for segmenting and smoothing 3D data may be used, but for the sake of brevity of this disclosure, no further detail is provided herein.

[0084] Figure 4 is a block diagram showing a computer system 400 in which various embodiments can be implemented. In some embodiments, the computer system 400 includes a bus 402 or other communication mechanism for communicating information and a hardware processor, or a plurality of processors 404, coupled to the bus 402 for processing information. The hardware processor 404 can be, for example, one or more general-purpose microprocessors.

[0085] In some embodiments, the computer system 400 also includes main memory 406, such as random access memory (RAM), cache, and / or other dynamic storage devices, coupled to the bus 402, for storing information and instructions executed by the processor 404. The main memory 406 may also be used to store temporary variables or other intermediate information during the execution of instructions to be executed by the processor 404. Once such instructions are stored in a storage medium accessed by the processor 404, the computer system 400 can be transformed into a special-purpose machine customized to perform the operations specified by the instructions. The main memory 406 may include instructions, for example, to analyze image information to determine the characteristics of coronary artery features (e.g., plaque, perivascular fat, and coronary arteries) and to generate a patient report containing information characterizing aspects of the patient's health related to the coronary arteries. For example, one or more metrics may be determined, and the metrics may include one or more of the following: feature slope / gradient, maximum density, minimum density, ratio of the slope of one feature to the slope of another feature, ratio of the maximum density of one feature to the maximum density of another feature, ratio of the minimum density of one feature to the minimum density of the same feature, or ratio of the minimum density of one feature to the maximum density of another feature.

[0086] In some embodiments, the computer system 400 further includes a read-only memory (ROM) 408 or other static storage device coupled to the bus 402 for storing static information and instructions for the processor 404. In some embodiments, a storage device 410, such as a magnetic disk, optical disk, or USB thumb drive (flash drive), is provided and coupled to the bus 402 for storing information and instructions.

[0087] The computer system 400 can be coupled via bus 402 to a display 412, such as a cathode ray tube (CRT) or LCD display (or touchscreen), for displaying information to the computer user. In some embodiments, an input device 414, including alphanumeric keys and other keys, is coupled to bus 402 to transmit information and command selections to the processor 404. Another type of user input device may include cursor control 416, such as a mouse, trackball, or cursor directional keys, for transmitting directional information and command selections to the processor 404 and controlling cursor movement on the display 412. In some embodiments, this input device typically has two degrees of freedom, with a first axis (e.g., x) and a second axis (e.g., y), allowing the device to specify a position in a plane. In some embodiments, the same directional information and command selection as cursor control may be implemented by receiving touches on a touchscreen without a cursor.

[0088] The computer system 400 may include a user interface module for implementing a GUI that can be stored in mass storage as computer executable program instructions executed by a computer device(s). The computer system 400 may further implement the techniques described herein using customized hardwired logic, one or more ASICs or FPGAs, firmware, and / or program logic that, in combination with the computer system, causes or programs the computer system 400 to become a special-purpose machine, as described below. According to some embodiments, the techniques described herein are executed by the computer system 400 in response to a processor(s) 404 that executes one or more sequences of one or more computer-readable program instructions contained in main memory 406. Such instructions may be read into main memory 406 from another storage medium, such as a storage device 410. In some embodiments, the execution of the instruction sequence contained in main memory 406 causes the processor(s) 404 to execute the process steps described herein. In some embodiments, hardwired circuitry may be used instead of or in combination with software instructions.

[0089] Various forms of computer-readable storage media may be involved in carrying one or more sequences of one or more computer-readable program instructions to the processor 404 for execution. For example, the instructions may initially be carried on a magnetic disk or solid-state drive of a remote computer. The remote computer may load the instructions into its dynamic memory and transmit them over a telephone line using a modem. A modem local to computer system 400 may receive the data over the telephone line and convert the data into an infrared signal using an infrared transmitter. An infrared detector may receive the data carried by the infrared signal, and appropriate circuitry may place the data on bus 402. Bus 402 carries the data to main memory 406, from which the processor 404 retrieves and executes the instructions. Instructions received by main memory 406 may optionally be stored in storage device 410 either before or after execution by processor 404.

[0090] In some embodiments, the computer system 400 also includes a communication interface 418 coupled to the bus 402. In some embodiments, the communication interface 418 provides a bidirectional data communication coupling to a network link 420 connected to a local network 422. For example, the communication interface 418 may be an Integrated Services Digital Network (ISDN) card, a cable modem, a satellite modem, or a modem for providing data communication connectivity to a corresponding type of telephone line. As another example, the communication interface 418 may be a local area network (LAN) card for providing data communication connectivity to a corresponding LAN (or WAN component for communicating with a WAN). Wireless links may also be implemented. In any such implementation, the communication interface 418 transmits and receives electrical, electromagnetic, or optical signals carrying digital data streams representing various types of information.

[0091] In some embodiments, the network link 420 typically provides data communication to other data devices via one or more networks. For example, the network link 420 may provide connection to a host computer 424 or data devices operated by an Internet service provider (ISP) 426 via a local network 422. In some embodiments, the ISP 426 provides data communication services through a global packet data communication network now commonly referred to as the “Internet” 428. Both the local network 422 and the Internet 428 can use electrical, electromagnetic, or optical signals to transmit digital data streams. Signals transmitted through various networks, signals on the network link 420, and signals via the communication interface 418 to and from the computer system 400 are examples of transmission media.

[0092] The computer system 400 can send messages and receive data, including program code, via the network, network link 420, and communication interface 418. In the internet example, server 430 might send request code for an application program via the internet 428, ISP 426, local network 422, and communication interface 418.

[0093] The received code is executed by processor 404 immediately upon receipt and / or stored in memory device 410 or other non-volatile memory for later execution.

[0094] Accordingly, in some embodiments, the computer system 105 includes a non-transient computer storage medium 410 configured to store at least patient image information. The computer system 105 may also include a non-transient computer storage medium that stores instructions for one or more processors 404 to perform a process (e.g., a method) for characterizing coronary plaque tissue data and perivascular tissue data using image data collected from computed tomography (CT) scans along blood vessels, wherein the image information includes radiation density values ​​of coronary plaque and perivascular tissue located adjacent to the coronary plaque. By executing instructions, in some embodiments, one or more processors 404 can quantify the radiation density in a region of coronary artery plaque in image data, quantify the radiation density in at least one region of corresponding perivascular tissue adjacent to the coronary artery plaque in image data, determine the gradient between the quantified radiation density value in the coronary artery plaque and the quantified radiation density value in the corresponding perivascular tissue, determine the ratio of the quantified radiation density value in the coronary artery plaque and the corresponding perivascular tissue, and characterize the coronary artery plaque by analyzing one or more of the gradient between the quantified radiation density value in the coronary artery plaque and the corresponding perivascular tissue, or the ratio between the radiation density value of the coronary artery plaque and the corresponding perivascular tissue.

[0095] Various embodiments of this disclosure may be systems, methods, and / or computer program products in any possible level of technical detail. A computer program product may include a computer-readable storage medium (or medium) having computer-readable program instructions thereon for causing a processor to perform an aspect of this disclosure. For example, the functions described herein may be executed by software instructions being executed by one or more hardware processors and / or any other suitable computing device, and / or in response to the software instructions being executed by one or more hardware processors and / or any other suitable computing device. The software instructions and / or other executable code may be read from the computer-readable storage medium (or medium).

[0096] A computer-readable storage medium can be a tangible device capable of holding and storing data and / or instructions used by an instruction execution device. A computer-readable storage medium can be, but is not limited to, electronic storage devices (including any volatile and / or non-volatile electronic storage devices), magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any suitable combination thereof. A non-exhaustive list of more specific examples of computer-readable storage media includes: portable computer diskettes, hard disks, solid-state drives, random-access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random-access memory (SRAM), portable compact disk read-only memory (CD-ROM), digital multipurpose disks (DVDs), memory sticks, floppy disks, mechanically encoded devices such as punch cards or grooved raised structures having instructions recorded thereon, and any suitable combination thereof. The computer-readable storage media used herein are not to be interpreted as transient signals themselves, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., light pulses passing through fiber optic cables), or electrical signals transmitted through wires.

[0097] The computer-readable program instructions described herein can be downloaded to the respective computing / processing device from a computer-readable storage medium or to an external computer or external storage device via a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network may include copper transmission cables, optical transmission fibers, wireless transmissions, routers, firewalls, switches, gateway computers, and / or edge servers. In some embodiments, a network adapter card or network interface within each computing / processing device receives computer-readable program instructions from the network and transfers the computer-readable program instructions for storage in a computer-readable storage medium within the respective computing / processing device.

[0098] Computer-readable program instructions for performing the operations of the Disclosure (also referred to herein, for example, as “code,” “instruction,” “module,” “application,” “software application,” etc.) may be either assembler instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, configuration data for integrated circuits, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smallalk and C++. Computer-readable program instructions are callable from other instructions or by themselves, and / or can be called in response to detected events or interrupts. Computer-readable program instructions configured to run on a computing device may be provided on a computer-readable storage medium and / or as a digital download (and may originally be stored in a compressed or installable format that requires installation, decompression, or decryption before execution), and subsequently stored on the computer-readable storage medium. Such computer-readable program instructions may be stored partially or entirely in the memory device of the executing computing device (e.g., a computer-readable storage medium) for execution by the computing device. Computer-readable program instructions may run completely on the user's computer (e.g., the executing computing device), partially on the user's computer as a standalone software package, partially on the user's computer and partially on a remote computer, or completely on a remote computer or server.In the latter scenario, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or it may be connected to an external computer (for example, via the Internet using an Internet service provider). In some embodiments, an electronic circuit including, for example, a programmable logic circuit, a field-programmable gate array (FPGA), or a programmable logic array (PLA) can be personalized by executing computer-readable program instructions using state information of computer-readable program instructions in order to perform aspects of the present disclosure.

[0099] Several aspects of the present disclosure are described herein with reference to flowcharts and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the present disclosure. It will be understood that each block in the flowcharts and / or block diagrams, as well as combinations of blocks in the flowcharts and / or block diagrams, can be implemented by computer-readable program instructions.

[0100] These computer-readable program instructions are provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device so that instructions executed via the processor of a computer or other programmable data processing device can create means for performing functions / actions specified in a flowchart and / or block diagram block or block, and can be used to manufacture a machine. These computer-readable program instructions can also be stored in a computer-readable storage medium that can instruct a computer, a programmable data processing device, and / or other device to function in a particular manner, and so a computer-readable storage medium having instructions stored therein comprises a product containing instructions that perform a manner of function / action specified in a flowchart(s) and / or block diagram(s).

[0101] Computer-readable program instructions can also be loaded into a computer, other programmable data processing device, or other device to cause a series of operational steps to be executed on the computer, other programmable device, or other device to generate a computer implementation process such that the instructions executed on the computer, other programmable device, or other device perform functions / actions specified in flowcharts and / or block diagram blocks or blocks. For example, instructions may initially be loaded onto a magnetic disk or solid-state drive of a remote computer. The remote computer can load the instructions and / or modules into its dynamic memory and transmit the instructions via telephone, cable, or optical lines using a modem. A modem local to the server computing system can receive data via telephone / cable / optical lines and, using a converter with appropriate circuitry, place the data onto a bus. The bus carries the data to memory, from which the processor retrieves and executes the instructions. Instructions received by memory can optionally be stored in a storage device (e.g., a solid-state drive) before or after execution by the computer processor.

[0102] The flowcharts and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of the systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of instructions comprising one or more executable instructions for implementing a specified logical function(s). In some alternative implementations, the functions described in a block may occur in a different order than shown in the figure. For example, two blocks shown consecutively may actually be executed substantially simultaneously, and depending on the functions involved, blocks may be executed in reverse order. Furthermore, in some embodiments, certain blocks may be omitted. Also, the methods and processes described herein are not limited to a specific order, and the blocks or states associated therewith may be executed in any other order as appropriate.

[0103] Based on identified features and / or quantified measurements, for example from the analysis of one or more medical images, in some embodiments, the system can be configured to use raw medical images to generate risk assessments and / or track the progression of plaque-based diseases or conditions, such as atherosclerosis, stenosis, ischemia, myocardial infarction, and / or major adverse cardiovascular events (MACE). As further described herein, in some embodiments, the system can perform risk assessments and / or track the progression of plaque-based diseases based on information from other patients. For example, features contained in a patient's medical image and patient information (e.g., age, sex, BMI, medication, blood pressure, heart rate, height, weight, race, whether the patient is a smoker or not, medical history, family history of the disease, etc.) can be compared or evaluated with features contained in the medical images and associated patient information of other patients (including outcomes after a period of time).

[0104] Furthermore, in some embodiments, the system may be configured to generate GUI visualizations of one or more identified features and / or quantified measurements, such as quantized color mappings of different features. In some embodiments, the systems, apparatus, and methods described herein are configured to utilize medical image-based processing to assess the risk of cardiovascular events, major adverse cardiovascular events (MACE), rapid plaque progression, and / or response or non-response to drug therapy and / or lifestyle changes and / or other treatments and / or invasive procedures for a subject. In particular, in some embodiments, the system may be configured to automatically and / or dynamically assess such health risks of a subject by analyzing only non-invasively obtained medical images. In some embodiments, one or more of the processes can be automated using artificial intelligence (AI) and / or machine learning (ML) algorithms. In some embodiments, one or more of the processes described herein can be performed in a reproducible manner within minutes. This is in contrast to existing means today that do not produce reproducible prognoses or assessments, require enormous amounts of time, and / or invasive procedures.

[0105] In some embodiments, image information comprising multiple images of a patient's coronary arteries and patient information / characteristics may be provided from one or more devices via a network to one or more servers of a processing system. The processing system is configured to generate coronary artery information using the multiple images of the patient's coronary arteries and to generate a two-dimensional and / or three-dimensional data representation of the patient's coronary arteries. The processing system then analyzes the data representation to generate a patient report documenting the patient's health status and risks associated with coronary plaque. The patient report may include images and graphs of the patient's arteries in terms of the type of coronary plaque within or near the coronary arteries. Machine learning techniques or other artificial intelligence techniques can be used to compare the data representation of the patient's coronary arteries with data representations of other patients (e.g., stored in a database) to determine additional information about the patient's health. In some embodiments, artificial intelligence may be trained using a dataset of data representations of other patients to identify correlations between data. For example, based on the state of specific plaques in the patient's coronary arteries, it may be possible to determine the likelihood of the patient having a heart attack or other adverse coronary artery effects. It may also be possible to determine additional information, for example, about the patient's CAD risk.

[0106] In some embodiments, the coronary artery plaque information of a patient being examined may be compared to, or referenced to, patients having one or more of the same or similar patient characteristics. For example, the patient being examined may be compared to patients having the same or similar characteristics of sex, age, BMI, medication, blood pressure, heart rate, weight, height, race, build, smoking, diabetes, hypertension, history of coronary artery disease, family history, and test results. Such comparisons can be performed by various means, such as machine learning and / or artificial intelligence techniques. In some examples, a neural network is used to compare the patient's coronary artery information with the coronary artery information of a large number (e.g., more than 10,000) other patients. A plaque risk assessment of the patient being examined can then be determined for such patients having similar patient and cardiac information.

[0107] In some embodiments, image information can be analyzed using deep learning (DL), machine learning (ML), and / or artificial intelligence (AI) methods. For example, this analysis may comprise image segmentation, feature extraction, and classification. In some embodiments, the ML method may comprise image feature extraction from raw data and image-based learning. In some embodiments, the ML method may accept input from a large training set and learn to ignore variability that could otherwise distort the results of the method. In some embodiments, DL may comprise a neural network (NN) with three or more layers that can improve the accuracy of the decisions. Advantageously, in some embodiments, DL eliminates the need to preprocess the data and can instead process the raw data. For example, a human can input a hierarchy of important features of coronary artery image information for the ML algorithm to make a decision, while the DL algorithm can determine which features are important and use these features to make the decision. Advantageously, in some embodiments, the DL algorithm can tune itself for accuracy and precision. In some embodiments, the ML and DL algorithms can perform supervised learning, unsupervised learning, and reinforcement learning.

[0108] In some embodiments, information can be analyzed in a manner similar to the high-level cognitive functions of the human mind, using NN approaches, particularly convolutional neural networks (CNNs) and recurrent convolutional neural networks (RCNNs). In some embodiments, the NN approach may involve training a large number of medical images to teach an object recognition system patterns in images that correlate with specific labels. In some embodiments, the CNN may comprise an NN in which the nodes of each layer are clustered, the clusters overlap, and each cluster feeds data to multiple nodes in the next layer. In some embodiments, the RCNN may comprise a CNN in which recurrent connections are incorporated into each convolutional layer. Advantageously, in some embodiments, the recurrent connections can make object recognition a dynamic process, even though the input is static.

[0109] In some embodiments, a vascular identification algorithm, a coronary artery identification algorithm, and / or plaque identification algorithm can be trained on multiple medical images in which one or more regions of blood vessels, coronary arteries, and / or plaques have been pre-identified. Based on such training, the system can be configured to automatically and / or dynamically identify the presence and / or parameters of blood vessels, coronary arteries, and / or plaques from raw medical images, for example, by using a CNN in some embodiments. In some embodiments, the system may be configured to utilize one or more AI and / or ML algorithms to identify and / or analyze blood vessels or plaques, derive one or more quantifiable indices and / or classifications, and / or generate a treatment plan. In some embodiments, the system may be configured to utilize AI and / or ML algorithms to identify intra-arterial, along-arterial, intra-arterial and / or extra-arterial regions of artery showing plaque accumulation. In some embodiments, input to the AI ​​and / or ML algorithms may include images of the patient, as well as patient information (or characteristics), such as age, sex, body mass index (BMI), medication, blood pressure, heart rate, height, weight, race, whether the patient is a smoker or non-smoker, body type, medical history, diabetes, hypertension, previous coronary artery disease (CAD), dietary habits, medication history, family history, information related to other previously collected image information, exercise habits, drinking habits, and one or more of the following: body type (e.g., "body type" or "physique," which can be based on a wide range of factors), medical history, history of diabetes, hypertension, coronary artery disease (CAD), dietary habits, medication history, family history of disease, information related to other previously collected image information, exercise habits, drinking habits, lifestyle information, test results, and / or similar. In examples where a neural network (NN) is used, the NN can be trained using information from multiple patients, and the information for each patient may include medical images and one or more patient characteristics.

[0110] In some embodiments, the system can be configured to automatically and / or dynamically identify one or more regions of plaque using image processing, utilizing one or more AI and / or ML algorithms. For example, in some embodiments, one or more AI and / or ML algorithms can be trained using a CNN on a set of medical images in which plaque regions have been identified, thereby enabling the AI ​​and / or ML algorithms to automatically identify plaque regions directly from the medical images. In some embodiments, the system can be configured to identify the vessel wall and lumen wall for each identified coronary artery in the medical images. In some embodiments, the system is then configured to determine the volume between the vessel wall and the lumen wall as plaque. In some embodiments, the system can be configured to identify plaque regions based on radiation density values ​​typically associated with plaque, for example, by setting a predetermined threshold or range of radiation density values ​​typically associated with plaque, with or without normalization using a normalization device.

[0111] In some embodiments, one or more vascular morphology parameters and / or plaque parameters may comprise quantified parameters derived from medical images. For example, in some embodiments, the system may be configured to determine one or more vascular morphology parameters and / or plaque parameters using AI and / or ML algorithms or other algorithms. As another example, in some embodiments, the system may be configured to determine one or more vascular morphology parameters, such as a classification of arterial remodeling by plaque, which may further include positive arterial remodeling, negative arterial remodeling, and / or intermediate arterial remodeling. In some embodiments, the classification of arterial remodeling is determined based on the ratio of the maximum vessel diameter in the plaque area to the normal reference vessel diameter in the same area, which can be retrieved from a normal database. In some embodiments, the system may be configured to classify arterial remodeling as positive if the ratio of the maximum vessel diameter in the plaque area to the normal reference vessel diameter in the same area is 1.1 or greater. In some embodiments, the system may be configured to classify arterial remodeling as negative if the ratio of the maximum vessel diameter in the plaque area to the normal reference vessel diameter is less than 0.95. In some embodiments, the system may be configured to classify arterial remodeling as intermediate when the ratio of the maximum vessel diameter in the plaque region to the normal reference vessel diameter is between 0.95 and 1.1.

[0112] In some embodiments, the system is configured to classify a subject's atherosclerosis into one or more high-risk, medium-risk, or low-risk categories based on quantified atherosclerosis. In some embodiments, the system is configured to classify a subject's atherosclerosis based on quantified atherosclerosis using AI, ML, and / or other algorithms. In some embodiments, the system is configured to classify a subject's atherosclerosis by combining and / or weighting one or more of the surface area-to-volume ratio, volume, heterogeneity index, and radioactivity density of one or more regions of the plaque.

[0113] In some embodiments, the system can be configured to automatically and / or dynamically identify one or more fatty regions, such as epicardial fat, in medical images using, for example, one or more AI and / or ML algorithms. In some embodiments, one or more AI and / or ML algorithms can be trained using a CNN on a set of medical images in which fatty regions have been identified, thereby enabling the AI ​​and / or ML algorithms to automatically identify fatty regions directly from medical images. In some embodiments, the system can be configured to identify fatty regions based on radiation density values ​​typically associated with fat, for example, by setting a predetermined threshold or range of radiation density values ​​typically associated with fat, with or without normalization using a normalization device.

[0114] In some embodiments, the system is configured to characterize changes in calcium score based on one or more plaque parameters derived from medical images, utilizing AI, ML, and / or other algorithms. For example, in some embodiments, the system may be configured to utilize AI and / or ML algorithms trained on a dataset of known medical images having identified plaque parameters combined with calcium scores, using a CNN. In some embodiments, the system may be configured to characterize changes in calcium score by accessing a known dataset of the same stored in a database. For example, the known dataset may include a dataset of calcium scores and / or medical images and / or changes in plaque parameters derived therefrom from other past subjects. In some embodiments, the system may be configured to characterize changes in calcium score and / or determine their causes on a vascular, segment, plaque, and / or subject basis.

[0115] In some embodiments, the systems disclosed herein can be used to dynamically and automatically determine the type, length, diameter, gauge, strength, and / or other arbitrary stent parameters required for a particular patient, based on processing of medical image data using, for example, AI, ML, and / or other algorithms.

[0116] In some embodiments, the system may be configured to generate patient-specific reports using AI and / or ML algorithms. In some embodiments, patient-specific reports may include documents, AR experiences, VR experiences, videos, and / or audio components.

[0117] Figure 5A is a block diagram illustrating an example of a process and / or system 800 (for ease of reference, both are referred to here as the "system") for identifying patient characteristics and / or risk information using AI / ML based on non-invasively acquired medical images and / or patient information of a patient. In some embodiments, the medical data of the current patient, including images and / or patient information, is first acquired and electronically stored in a medical data storage device 816 (e.g., cloud storage, hard disk, etc.). In some embodiments, the system 800 acquires medical images and / or patient information 818 from the medical data storage device 816, preprocesses it if necessary, and reformats it as needed for further processing, for example. The system 800 can also obtain a training set 822 of medical images and / or patient information from a stored dataset 820 of medical images and / or information of other patients (e.g., hundreds, thousands, tens of thousands, or hundreds of thousands of other patients). Medical images and information of other patients can be used to train the AI / ML algorithm 824 before processing the medical images and / or patient information 818 of the current patient, as will be further described with reference to Figures 5C and 5D. In some embodiments, the AI / ML algorithm 824 may include one or more neural networks (NNs), as will be described with reference to the exemplary NN illustrated in Figure 6, for example. The ML / AI 824 processes the medical images and / or patient information 818 of the current patient and generates an output of identified features and / or risk information 826 of the current patient.

[0118] Figure 5B is a schematic diagram showing an example of an NN812 that makes a determination 814 regarding the characteristics of the (current) patient based on an input including a medical image 802. In some embodiments, the NN812 can be configured to receive other inputs 804. In some embodiments, the other inputs 804 may be medical images of other patients. In some embodiments, the other inputs 804 may be the medical history of other patients. In some embodiments, the other inputs 804 may be the medical history of the (current) patient. The NN812 may include an input layer 806. In some embodiments, the NN812 may be configured to present a training pattern to the input layer 806. In some embodiments, the NN812 may include one or more hidden layers 808. In some embodiments, the input layer 806 may provide signals to the hidden layer 808, and the hidden layer 808 may receive signals from the input layer 806. In some embodiments, the hidden layer 808 may pass signals to an output layer 810. In some embodiments, one or more hidden layers 808 may consist of convolutional layers (with weighted neurons / nodes connected by weights, where weights correspond to the strength of the connections between neurons), pooling layers, fully connected layers, and / or normalization layers. In some embodiments, NN812 may consist of pooling layers that connect the outputs of neuron clusters in one layer to a single neuron in the next layer. In some embodiments, max pooling and / or average pooling may be utilized. In some embodiments, max pooling may utilize the maximum value from each of the neuron clusters in the previous layer. In some embodiments, backpropagation may be utilized, and the corresponding neural network weights may be adjusted to minimize or reduce the error. In some embodiments, the loss function may consist of a binary cross-entropy loss function.

[0119] In some embodiments, the NN812 may include an output layer 810. In some embodiments, the output layer 810 may receive signals from the hidden layer 808. In some embodiments, the output layer may generate a decision 814. In some embodiments, the NN812 may make a decision 814 regarding patient characteristics. In some embodiments, the decision 814 may include a set of plaque features. In some embodiments, the decision 814 may include the patient's CAD risk.

[0120] Figure 5C shows an example of a process in a flowchart for training an artificial intelligence or machine learning model. Process 828 can be executed on a computer system. Various embodiments of such a process for training an AI or ML model may include additional features and / or exclude certain illustrated features (for example, if a transformed / preprocessed dataset is received so that “Apply transformations” in block 832 does not need to be performed).

[0121] As shown in the example in Figure 5C, in block 830, the system receives the dataset. In block 832, one or more transformations may be performed on the data in the dataset. In one example, the data may require transformations to conform to an expected input format, such as a date format, units (e.g., pounds versus kilograms, Celsius versus Fahrenheit, inches versus centimeters, etc.), or to be in a consistent format. In some embodiments, the data may undergo transformations for use in training AI or ML algorithms; for example, categorical data may be encoded in a particular way. In some embodiments, nominal data may be encoded using one-hot encoding, binary encoding, feature hashing, or other suitable encoding methods. In some embodiments, ordinal data may be encoded using ordinal coding, polynomial coding, Helmert coding, etc. In some embodiments, numerical data may be normalized, for example, by scaling the data to a maximum value of 1 and a minimum value of 0 or -1. In some embodiments, the dataset may include images, which may undergo resizing, orientation, color correction, tilt correction, color space transformation, etc. These are merely examples, and those skilled in the art will readily understand that other transformations are possible.

[0122] In block 834, the system can create a training dataset, a tuning dataset, and a test / validation dataset from the received dataset. In some embodiments, the training dataset 836 may be used during training to determine features for forming a model that can be used for prediction, classification, etc. In some embodiments, the tuning dataset 838 may be used to select the final model (e.g., final model weights) and to prevent or correct overfitting that may occur during training with the training dataset 836, otherwise it may lead to poor generalization of the model. In some embodiments, the test dataset 840 may be used after training and tuning to evaluate the model. For example, in some embodiments, the test dataset 840 may be used to check whether the model is overfitting the training dataset. For example, if iterative training is used, overfitting may be indicated by a continuous improvement in the model's performance on the training data (e.g., a continuous improvement in the loss function or error), while performance on the test dataset may improve for a certain period or number of training iterations, but then begin to decline.

[0123] In some embodiments, the system may train a model in block 842 using the training dataset 836 in the training loop 856. In some embodiments, training may be performed in a supervised, unsupervised, or partially supervised manner. In some embodiments of this disclosure, supervised training may be used. In 844, in some embodiments, the system may evaluate the model according to one or more evaluation criteria. For example, in some embodiments, evaluation may include determining the extent to which the model can determine image transformations to account for changes in image acquisition parameters. In other embodiments, evaluation may include determining how well the model can identify thin capsular fibromas. Other applications of the AI / ML model are described herein, and different evaluation criteria may be used for different applications. In 846, in some embodiments, the system may determine whether the model meets one or more evaluation criteria. In some embodiments, if the model fails the evaluation, the system may tune the model in 848 using the tuning dataset 838 and repeat training 842 and evaluation 844 until the model passes the evaluation in 846. In some embodiments, if the model passes the evaluation in 846, the system can exit the model training loop 856. In some embodiments, the test dataset 836 is run through the trained model 842, and in block 844, the system can evaluate the results. In some embodiments, if the evaluation fails, in block 846, the system can re-enter the training loop 856 for additional training and tuning. If the model passes, the system stops the learning process and the trained model 850 is obtained. In some embodiments, the training process can be modified. For example, in some embodiments, the system may not use the tuning dataset 838. In some embodiments, the system may not use the test dataset 840.

[0124] In some embodiments, testing can be performed within the training loop 856, and training can be stopped when the model's performance on the test data stops improving or begins to decrease. For example, training can be stopped to avoid overfitting the model to the training data. In some embodiments, training can be stopped after a defined number of iterations.

[0125] Figure 5D shows an example of the process of training and using an AI / ML model. In some embodiments, the training data store 858 can store data for training the model. For example, in some embodiments, the training data store 858 can store medical images of a patient and / or information about the patient's physiology, such as weight and BMI. In block 860, in some embodiments, the system may be configured to prepare training data if training data has not been previously prepared for use in training a model. In some embodiments, as briefly described above, preparing training data may include performing one or more normalization procedures such as unit conversions (e.g., between Fahrenheit and Celsius, between inches and centimeters, between pounds and kilograms), conversion of dates to standard formats, conversion of times to standard formats, standardization procedures, and image processing (e.g., size, orientation, color space, etc.). In some embodiments, it may be desirable to exclude certain data because adding data would consume more computational resources and take longer to train the model. However, in some embodiments, exclusion may be undesirable because the model may not accurately consider the effect of changes in image acquisition parameters on the resulting outcomes. In block 862, the system can extract features from the training data, and in block 864, the system can train the model using the training data to generate model 866. In block 868, in some embodiments, the system can evaluate the model to determine whether it passes one or more criteria. In some embodiments, if the model fails at decision point 870, the system can perform additional training. In some embodiments, if the model passes at decision point 870, the system can make the trained model 872 available.

[0126] In some embodiments, the trained model 872 can be used to evaluate a specific patient. The input data 874 may relate to a specific input for which the output of the trained model 872 is desired. In block 876, the system can prepare the input data 874 as described above, for example, in relation to stored training data. In some embodiments, in block 878, the system can extract features from the prepared user data. In some embodiments, the system may be configured to feed the extracted features to the trained model 872 to produce a result 880.

[0127] In some embodiments, the input data 874, the results 880, and / or other information can be used to train the model. In block 882, in some embodiments, the system can prepare the input data 874 and the results 880 for use in training the model 872. In some embodiments, the system can store the prepared data in the training data store 858. In some embodiments, the prepared data can be stored in an additional or alternative database or data store. In some embodiments, the system can retrain the model periodically, continuously, or whenever the operator indicates to the system that the model needs to be retrained. Thus, in some embodiments, the trained model 872 can evolve over time, and as a result, the model's performance (e.g., improved predictive ability, improved classification ability, etc.) can improve over time.

[0128] The dataset used for training or testing may include, for example, CT images, computed tomography angiography (CCTA) images, extracted vessels, image acquisition parameters, patient information (e.g., patient identifier, patient weight, patient body mass index, etc.), and / or other relevant information.

[0129] In some embodiments, a machine learning model can be trained using supervised learning, where the training data includes input data (e.g., images, patient information, etc.) as inputs and desired outputs (e.g., classification, stratification, labeling, etc.), such as risk assessment, labeled coronary artery trees. A representation of the input data can be provided to the model. The output from the model can be compared to the desired output. For example, in a classification model, the desired output may be the true classification of the input, which can be compared to the classification determined by the model. In some embodiments, based on the comparison, the model can be modified by changing the weights associated with the nodes of the neural network or the parameters of the functions used at each node of the neural network (e.g., applying a loss function). The model can be modified until it produces the desired output with the desired accuracy.

[0130] [Computer System] In some embodiments, the systems, processes, and methods described herein are implemented using a computer system such as that illustrated in Figure 6. The exemplary computer system 928 communicates with one or more computer systems 946 and / or one or more data sources 948 via one or more networks 944. Figure 5A shows one embodiment of computer system 928, but it is recognized that the functions provided in the components and modules of computer system 928 can be combined into fewer components and modules, or further separated into additional components and modules.

[0131] The computer system 928 may include a plaque analysis module 940 that performs the functions, methods, actions, and / or processes described herein. The plaque analysis module 940 is executed on the computer system 928 by a central processing unit 306, which will be discussed further below.

[0132] Generally, as used herein, the term “module” refers to logic embodied in hardware or firmware, or a set of software instructions with inputs and outputs. Modules are written in programming languages ​​such as Java®, C#, C, and C++. Software modules can be compiled or linked into executable programs, installed into dynamic-link libraries, or written in interpreted languages ​​such as JavaScript, BASIC, Perl, Lua, PHP, and Python. Software modules can be called from other modules or from themselves, and / or in response to detected events or interrupts. Modules implemented in hardware may include connected logic units such as gates and flip-flops, and / or programmable units such as programmable gate arrays and processors.

[0133] In general, the modules described herein refer to logical modules that, despite their physical configuration and storage, can be combined with other modules or divided into submodules. Modules can be executed by one or more computer systems, stored on or within any suitable computer-readable medium, or implemented entirely or partially in specially designed hardware or firmware. Not all calculations, analyses, and / or optimizations require the use of computer systems, however any of the methods, calculations, processes, or analyses described herein can be facilitated by the use of computers. Furthermore, in some embodiments, the process blocks described herein can be modified, rearranged, combined, and / or omitted.

[0134] Computer system 928 includes one or more processing units (CPUs) 932, which may be provided by a microprocessor. Computer system 928 further includes physical memory 936, such as random access memory (RAM) for temporary storage of information and read-only memory (ROM) for persistent storage of information, and mass storage devices 930, such as backing stores, hard drives, spinning magnetic disks, solid-state disks (SSDs), flash memory, phase-change memory (PCM), 3D XPoint memory, diskettes, or optical media storage devices. Alternatively, mass storage devices may be implemented in a server array. Typically, the components of computer system 928 are connected to the computer using a standards-based bus system. The bus system can be implemented using various protocols, such as PCI (Peripheral Component Interconnect), PCI Express, Microchannel, SCSI, ISA (Industrial Standard Architecture), and EISA (Extended ISA) architectures.

[0135] The computer system 928 includes one or more input / output (I / O) devices and interfaces 938, such as a keyboard, mouse, touchpad, and printer. The I / O devices and interfaces 938 may include one or more display devices, such as monitors, that enable the visual presentation of data to the user. More specifically, the display devices may provide, for example, the presentation of a GUI as application software data and the presentation of multimedia presentations. The I / O devices and interfaces 938 may also provide communication interfaces to various external devices. The computer system 928 may also include one or more multimedia devices 934, such as speakers, video cards, graphics accelerators, and microphones.

[0136] [Computer system equipment / Operating system] Computer system 928 can run on a variety of computer devices, including servers, Windows servers, Structure Query Language servers, Unix® servers, personal computers, and laptop computers. In other embodiments, computer system 928 can run on cluster computer systems, mainframe computer systems, and / or other computing systems suitable for controlling and / or communicating with large databases, performing high-volume transaction processing, and generating reports from large databases. Computer system 928 is generally controlled and tuned by operating system software, such as z / OS, Windows, Linux®, UNIX®, BSD, SunOS, Solaris, macOS®, iOS, iPadOS®, or other compatible operating systems, including proprietary and / or open-source operating systems. The operating system controls and schedules the execution of computer processes, performs memory management, provides file systems, networking, and I / O services, and provides user interfaces such as graphical user interfaces (GUIs).

[0137] [network] The computer system 928 shown in Figure 6 is connected to a network 944, such as a LAN, WAN, or the Internet, via a communication link 942 (wired, wireless, or a combination thereof). Network 944 communicates with various computing devices and / or other electronic devices. Network 944 communicates with one or more computer systems 946 and one or more data sources 948. The plaque analysis module 914 can access, or is able to access, computer systems 946 and / or data sources 948 via a web-enabled user access point. The connection can be a direct physical connection, a virtual connection, or other connection type. The web-enabled user access point may comprise a browser module that presents data using text, graphics, audio, video, and other media, and enables interaction with the data via network 944.

[0138] The output module can be implemented as a fully addressable display such as a cathode ray tube (CRT), liquid crystal display (LCD), plasma display, or other types and / or combinations of displays. The output module can be implemented to communicate with the input device 938, and they also include software with a suitable interface that allows the user to access data through the use of stylized screen elements such as menus, windows, dialog boxes, toolbars, and controls (e.g., radio buttons, checkboxes, slide scales, etc.). Furthermore, the output module can communicate with a set of input / output devices to receive signals from the user.

[0139] [Other systems] The computer system 928 may include one or more internal and / or external data sources (e.g., data source 948). In some embodiments, one or more of the aforementioned data repositories and data sources may be implemented using relational databases such as DB2, Sybase, Oracle, CodeBase, and Microsoft® SQL Server, as well as other types of databases such as flat file databases, entity-relational databases, object-oriented databases, and / or record-based databases.

[0140] The computer system 928 can also access one or more databases 948. Databases 948 can be stored in a database or data repository. The computer system 928 can access one or more databases 948 via the network 944, or directly via the I / O devices and interface 938 to access a database or data repository. A data repository containing one or more databases 948 can reside within the computer system 928.

[0141] [URL and cookies] In some embodiments, including any of the embodiments disclosed herein (above or below), one or more features of the systems, methods, and apparatus described herein may utilize URLs and / or cookies, for example, to store and / or transmit data or user information. A uniform resource locator (URL) may include a web address and / or a reference to a web resource stored in a database and / or server. A URL may specify the location of a resource on a computer and / or computer network. A URL may include a mechanism for retrieving a network resource. The source of a network resource may receive a URL, locate the web resource, and send the web resource back to the requester. A URL can be translated into an IP address, and a Domain Name System (DNS) may look up URLs and their corresponding IP addresses. A URL may be a reference to a web page, file transfer, email, database access, or other application. A URL may include a path, domain name, file extension, host name, query, fragment, scheme, protocol identifier, port number, username, password, flag, object, resource name, and / or sequence of characters identifying them. The systems disclosed herein can generate, receive, send, apply, parse, serialize, render, and / or perform actions on URLs.

[0142] Cookies, also known as HTTP cookies, web cookies, internet cookies, or browser cookies, can contain data transmitted from a website and / or stored on a user's computer. This data may be stored by the user's web browser while the user is browsing. Cookies can contain useful information for websites to remember previous browsing information, such as online store shopping carts, button clicks, login information, and / or records of previously visited web pages or network resources. Cookies can also contain information entered by the user, such as name, address, password, and credit card information. Cookies can also perform computer functions. For example, authentication cookies can be used by applications (e.g., web browsers) to identify whether a user is already logged in (e.g., to a website). Cookie data can be encrypted to provide security to consumers. Tracking cookies can be used to compile an individual's past browsing history. The systems disclosed herein can generate and use cookies and access personal data. The systems can also generate and use JSON web tokens to store authenticity information, HTTP authentication as an authentication protocol, IP addresses, URLs, etc., for tracking session or identity information.

[0143] [Normalization of medical images for plaque and vascular analysis - 8006] This specification describes systems, methods, and apparatus for normalizing medical images for coronary plaque analysis, vascular analysis, and / or similar applications. In some embodiments, the systems, methods, and apparatus described herein may be configured to utilize one or more AI / ML models for normalizing medical images.

[0144] Cardiac patients, such as those with heart disease, may undergo sequential imaging procedures (e.g., annual or quarterly imaging). These imaging procedures may include, for example, computed tomography (CT) scans and coronary angiography (CCTA) scans. Comparing images acquired over time can be important, for example, to determine whether the patient's condition is stable, improving, or worsening. However, comparing images over time can be very difficult. For example, different imaging devices may be used, and different acquisition parameters may be used during image acquisition (e.g., different integration times, X-ray intensity, X-ray energy, peak kilovolts (kVp), current (mA), etc.). In some cases, the patient's physiological functions may change. For example, weight gain or loss in a patient can affect X-ray attenuation by tissue. In some cases, the patient's hydration level can also affect X-ray attenuation. In addition, tissues are generally not rigid and may move due to the patient's position, respiration, heart rate, etc. Therefore, co-registering two consecutive imaging scans and / or determining clinically significant differences between the two scans (e.g., differences arising from different imaging conditions or unrelated changes in the patient's physiology) can be difficult. Without reliable criteria or other methods to ensure meaningful comparisons between images, determining changes such as plaque, lumen measurements, and lumen composition can be challenging.

[0145] One approach to normalizing images acquired at different times is to use a physical calibration device during the imaging procedure. A physical calibration device can include one or more standards, such as water that can be imaged with the same settings as the patient's image. In some cases, the standard can include a sticker or patch worn by the patient during the imaging procedure. In some cases, images of the physical calibration device can be compared between imaging procedures to inform how the patient image should be adjusted to account for differences in imaging conditions. However, such an approach has several drawbacks. For example, imaging a physical calibration device can add time to the imaging procedure and consume additional memory. In some cases, time may have to be spent adjusting the physical calibration device image to determine how to adjust the patient image. If a sticker is applied to the patient's body, the placement of the sticker may add extra time or steps to the imaging procedure, and different placements of the sticker may affect the apparent electromagnetic properties of the sticker.

[0146] The imaging results can depend not only on the patient's physiology but also on various imaging parameters. For example, the imaging results can vary depending on the helical CT scan method (e.g., single source, dual source, multi-source, rapid switching, high-speed pitch helical), the type of detector (e.g., energy integrating detector or photon counting detector), the type of energy used (e.g., single energy or dual energy), the current (mA) passing through the X-ray generator, peak kilovolts (kVp), image noise, signal level, signal-to-noise ratio, contrast effect, contrast-to-noise ratio, etc.

[0147] According to several embodiments, an image normalization algorithm can be provided. For example, in some embodiments, a database, data store, or other data source may contain a large number of previously acquired images. In some embodiments, each patient represented in the data store may have undergone one or more imaging procedures using different image acquisition parameters. In some embodiments, polynomial regression can be performed to determine how to normalize a second scan based on a first scan so that a direct comparison can be made between a first scan and a second scan.

[0148] As described herein, there may be numerous variables that can affect the image, including the settings of the imaging device (e.g., a CT scanner) and / or changes in the patient's physiology over time. The image may change linearly due to some variables, but non-linearly due to others. Therefore, performing polynomial regression to develop an image processing algorithm is difficult, or even impossible. In some embodiments, an AI / ML model can be used to generate an image processing algorithm and / or to process the image. For example, in some embodiments, an AI / ML model can be trained by providing the model with a set of images having different known image acquisition parameters and / or patient physiological parameters, and training the model to determine the modifications that should be made to the images to account for differences in one or more image acquisition parameters and / or one or more differences in patient physiology.

[0149] In some embodiments, scan normalization can be performed using a single model for all imaging systems (e.g., all CT scanners). However, such an approach may require a relatively large number of samples to obtain reliable ground truth information that can be broadly applied to different CT scanners or other imaging devices. In addition, a generalized model for all CT scanners or other imaging devices may be relatively more complex than a model for a specific CT scanner, a specific model of a CT scanner, etc. In some embodiments, the normalization model for individual CT scanners may be set at onboarding or configured in other ways. In some embodiments, the normalization model may be shared by a specific model or group of CT scanners (e.g., multiple models of CT scanners may have the same or very similar imaging hardware). However, in some embodiments, even individual scanners of the same model may exhibit somewhat different performance. Therefore, it may be important to ensure that the normalization model accurately normalizes images for a particular CT scanner. In some embodiments, the normalization model can be tailored for a specific CT scanner. In some embodiments, the normalization model may be periodically adjusted to maintain and / or improve the performance of the normalization model, whether it is specific to a particular scanner, specific to a scanner model, or not.

[0150] The challenges of normalization are not necessarily limited to image acquisition conditions or patient physiology. For example, in some cases, the image reconstruction algorithm may affect the apparent intensity depending on the definition of voxels, the type of iterative reconstruction, etc. In some cases, the noise level in the image may affect normalization. In some embodiments, the reconstruction approach can be standardized, which can ensure relatively consistent results for the image reconstruction algorithm.

[0151] In some embodiments, image normalization can be performed at a global level, for example, regardless of anatomical features present in the image. In some embodiments, image normalization can be performed at a more local level, which can help account for local effects such as blooming.

[0152] There can be many different CT scanners, many different imaging parameters, etc. In some embodiments, images from actual imaging procedures can be used in training AI / ML models for image normalization and / or in determining parameters for polynomial regression. In some embodiments, images can be simulated using different imaging parameters, either additionally or alternatively. Using simulated images can greatly increase the number of usable images, the diversity of usable imaging parameters, etc., and in some cases may lead to improvements in the model or polynomial regression.

[0153] In some embodiments, the system may be configured to present a default normalization to the user. In some embodiments, the system may provide an interface, configuration file, etc., that allows the user to adjust the normalization settings, thereby enabling the user to optimize and / or customize image normalization using expert judgment.

[0154] [Risk stratification for coronary artery disease / *8007* / ] Furthermore, this specification discloses systems, methods, and apparatus for image-based coronary artery disease (CAD) risk stratification. In some embodiments, the systems, methods, and apparatus described herein may be configured to integrate one or more inputs of stenosis, atherosclerosis, ischemia, or a combination thereof to determine and / or facilitate the determination of a subject's CAD risk stratification. In some embodiments, additional information such as information on placement, diameter, length, stent type, etc., may be input. In some embodiments, the systems, methods, and apparatus described herein may utilize one or more AI / ML algorithms to combine and / or consider one or more inputs of stenosis, atherosclerosis, ischemia, additional information, or any combination thereof to determine and / or facilitate the determination of a subject's CAD risk stratification. In some embodiments, the systems, methods, and apparatus described herein may be configured to provide a prognosis of a subject's CAD risk and / or the likelihood of the subject experiencing a major adverse cardiovascular event (MACE). In some embodiments, the approaches described herein can be used to determine the predicted prognosis based on one or more possible treatments, such as stent placement, drug therapy, dietary changes, and exercise changes. In some embodiments, the approaches described herein can be used to determine the prognosis over time. For example, a physician can consider the short-term and long-term benefits and risks of a particular treatment.

[0155] Current solutions and / or technologies for risk stratification of CAD do not take into account the overall situation, including but not limited to stenosis, atherosclerosis, ischemia, or any combination thereof. To address such technical shortcomings, some embodiments of the systems, methods, and apparatus described herein are configured to integrate and / or consider one or more inputs of stenosis, atherosclerosis, ischemia, or a combination thereof, and / or additional inputs such as potential treatments, thereby providing improved and / or more accurate risk stratification and / or staging of CAD in subjects.

[0156] In some embodiments, the systems, methods, and apparatus described herein may be configured for use in screening symptomatic and / or asymptomatic subjects for risk stratification of CAD. In some embodiments, the systems, methods, and apparatus described herein may be configured to provide MACE prognostic output. In some embodiments, the systems, methods, and apparatus described herein may be configured to provide MACE prognostic output better than an atherosclerotic cardiovascular disease (ASCVD) risk score or ASCVD risk calculator. In some embodiments, the systems, methods, and apparatus described herein may be configured to improve upon the ASCVD risk score or risk calculator in terms of positive predictive value (PPV) for disease or CAD stage, and / or provide MACE prognostic output better than the ASCVD risk score or risk calculator. In some embodiments, the systems, methods, and apparatus described herein may be configured to provide MACE prognostic output with, for example, a stage-based hazard ratio having a confidence interval of 95% or more. The prognostic outputs of MACE obtained through the approaches described herein may show improved and / or superior integrated discriminant improvement (IDI) indices compared to ASCVD risk scores or risk calculators. In some embodiments, the systems, methods, and apparatus described herein may be configured to output CAD staging systems and / or CAD risk stratification systems that improve patient outcomes. In some embodiments, the systems, methods, and apparatus described herein may be configured to output CAD staging systems and / or CAD risk stratification systems that direct guideline-oriented treatments tailored to underlying disease risk (e.g., CAD risk) to prevent or reduce the likelihood of MACE.

[0157] In some embodiments, the systems, methods, and apparatus described herein are configured to generate CAD stages for a subject. In some embodiments, the systems, methods, and apparatus described herein are configured to classify and / or generate CAD stages for a subject. In some embodiments, the CAD stages may be ordinal, for example, such that the system determines that the subject has a CAD risk corresponding to stage 0, stage 1, stage 2, or stage 3. In some embodiments, the system determines that the total plaque volume or other plaque volume of the subject is approximately 0 mm 3 The system may be configured to determine that a subject has a stage 0 CAD risk if the following conditions are met. In some embodiments, the system determines the subject's total plaque volume or other plaque volume is approximately 1 to approximately 250 mm 3 The system may be configured to determine that the subject has a Stage 1 CAD risk if the following conditions are met. In some embodiments, the system determines that the subject's total plaque volume or other plaque volume is approximately 251 to approximately 750 mm 3 If this is the case, the system may be configured to determine that the subject has a Stage 2 CAD risk. In some embodiments, the system determines that the subject's total plaque volume or other plaque volume is approximately 750 mm 3 The system may be configured to determine that a subject has a Stage 3 CAD risk if the threshold exceeds a certain value. In some embodiments, the systems, apparatus, and methods described herein are configured to output a continuous CAD risk assessment of a subject. That is, in some embodiments, the system may be configured to generate a numerical or value on a continuous scale that represents the subject's CAD risk. In some embodiments, the end user (e.g., a physician) can adjust the thresholds that distinguish different CAD stages.

[0158] In some embodiments, the systems, methods, and apparatus described herein may be configured to increase, decrease, jump up, or otherwise modify a previously generated CAD risk assessment or stratification of a subject based, for example, on plaque volume or total plaque volume. For example, in some embodiments, the system may be configured to increase, jump up, or otherwise modify a subject's CAD risk assessment or stage if severe stenosis is found in a specific location or coronary artery, such as the left main coronary artery (LMCA), right coronary artery (RCA), right marginal artery, left coronary artery, circumflex artery, left obtuse-angled marginal artery, left anterior descending artery, proximal left anterior descending artery, or diagonal artery. In particular, in some embodiments, the system may be configured to increase, jump up, or otherwise modify a subject's CAD risk assessment or stage if severe stenosis is found in the left main coronary artery or proximal LAD. In some embodiments, the system may be configured to increase, jump up, or otherwise modify the CAD risk assessment or stage for a subject if ischemia or the possibility of ischemia is found in one or more of the aforementioned coronary arteries.

[0159] In some embodiments, the system may be configured to change the CAD risk assessment or stratification up or down, increment or decrement, or otherwise modify it by 1, 2, 3, or one or more steps on a continuous and / or ordinal scale. In some embodiments, the system is configured to ascend only a limited number of steps, e.g., one or two steps, regardless of the number of stenoses and / or ischemias found.

[0160] In some embodiments, the systems, methods, and apparatus described herein are configured to combine one or more of atherosclerosis, plaque, stenosis, and / or ischemia to generate a CAD risk assessment or stratification or classification of a subject. In some embodiments, the system may be configured to combine one or more of atherosclerosis, plaque, stenosis, and / or ischemia by utilizing one or more ratios, exponents, arbitrary mathematical functions, logarithmic equations, algorithmic equations, and / or machine learning. In some embodiments, the generated CAD risk assessment or stratification may comprise one or more values ​​on a continuous scale, staging, or classification.

[0161] In some embodiments, systems, methods, and apparatus may be configured to utilize any one or more plaque characteristics discussed herein, including, for example, the amount or type of plaque, location, remodeling, degree of embedding in other types of plaque, distance from plaque to the blood vessel or lumen wall, or shape, in generating CAD risk stratification.

[0162] In some embodiments, the systems, methods, and apparatus of this specification may be configured to take stenosis into consideration when generating CAD risk assessments or stratifications. In some embodiments, the system may be configured to utilize stenosis in a binary form, for example, as either presence or absence of stenosis. In some embodiments, the system may be configured to utilize the presence or absence of severe and / or non-severe stenosis, for example, with a predetermined cutoff value, as input. In some embodiments, the system may be configured to utilize the severity of stenosis on a continuous, stepwise, or ordinal scale as input.

[0163] In some embodiments, the systems, methods, and apparatus of this specification may be configured to take ischemia into consideration when generating CAD risk assessments or stratifications. In some embodiments, the system may be configured to utilize ischemia in a binary form, for example, as the presence or absence of ischemia or the likelihood of the presence or absence of ischemia (e.g., likely or unlikely). In some embodiments, the system may be configured to utilize, for example, the presence or absence of severe and / or non-severe ischemia with a predetermined cutoff value as input. In some embodiments, the system may be configured to utilize the severity of ischemia on a continuous, stepwise, and / or ordinal scale as input, for example, using AI-based fractional flow reserve or AI-FFR.

[0164] In some embodiments, the systems, methods, and apparatus described herein may be configured to combine analysis of plaque, stenosis, ischemia, and / or pericoronary adipose tissue when generating risk stratification or assessment of CAD in subjects. In some embodiments, the system may also be configured to consider the percentage of attenuated myocardium, perfusion, and / or endothelial wall shear stress when generating risk stratification or assessment of CAD in subjects. In some embodiments, additional factors may be considered, such as treatments that may include lifestyle changes, medications, and / or surgical interventions.

[0165] In some embodiments, the systems, methods, and apparatus described herein may be configured to raise or lower, increase or decrease, or otherwise modify CAD risk stratification or assessment based solely on image analysis. In some embodiments, the systems, methods, and apparatus described herein may be configured to raise or lower, increase or decrease, or otherwise modify CAD risk stratification or assessment based on image analysis and in combination with one or more other factors, such as laboratory tests, subject symptoms, biometrics, risk factors, and / or molecular diagnostics.

[0166] In some embodiments, the systems, methods, and apparatus described herein may be configured to generate CAD risk assessments or stratifications based on an ordinal or continuous scale. In some embodiments, the systems, methods, and apparatus described herein may be configured to generate CAD risk assessments on an absolute or relative scale compared to normal risk levels in a reference population, such as those determined at least in part on similar demographics, such as the sex or age of the subjects.

[0167] In some embodiments, the systems, methods, and apparatus described herein may be configured to generate a CAD risk assessment or stratification linked to a specific time axis of MACE. For example, the CAD risk assessment may be based on the likelihood of MACE occurring within a specific time frame. In some embodiments, the time frame may be within a range defined by about 3 months, about 6 months, about 1 year, about 1.5 years, about 2 years, about 2.5 years, about 3 years, about 3.5 years, about 4 years, about 4.5 years, about 5 years, about 5.5 years, about 6 years, about 6.5 years, about 7 years, about 7.5 years, about 8 years, about 8.5 years, about 9 years, about 9.5 years, about 10 years, and / or two of the aforementioned values.

[0168] In some embodiments, the risk stratification model may take into account various additional and / or alternative information for determining risk stratification. For example, in some embodiments, the system may be configured to use perfusion data (e.g., obtained using MRI) to determine ischemia. In some embodiments, the system may be configured to use myocardial blood flow (e.g., determined using SPECT) to determine ischemia. In some embodiments, the system may be configured to use intravascular ischemia data, such as determined via invasive FFR, which can provide information related to stenosis. In some embodiments, intravascular ischemia data can provide information related to plaque. In some embodiments, the AI / ML model may be provided with anatomical inputs to determine physiology, which can provide information about stenosis, plaque, ischemia, or any combination thereof. In some embodiments, a machine learning model can be trained to predict MACE. For example, the model can be trained using data on type I myocardial infarction, type II myocardial infarction, stroke, etc.

[0169] In some embodiments, the prognosis may vary depending on any treatment provided to the patient, such as changes in diet or exercise, medication, or local interventions such as stent placement. In some embodiments, the prognosis may vary depending on the length of the stent, the diameter of the stent, and the placement of the stent. For example, as will be described in more detail herein, the selection and placement of the stent can have a significant impact on the patient's outcome. For instance, if the stent is too short, restenosis is more likely to occur. If the stent is too wide compared to the diameter of the vessel in which it is implanted, the vessel is more likely to rupture. Improper stent placement can lead to adverse effects such as in-stent restenosis.

[0170] In some embodiments, the intervening physician may utilize the systems, methods, and apparatus described herein to assist in stent selection and / or in planning the surgical procedure for stent placement. In some embodiments, the intervening physician may utilize the approaches described herein to determine prognosis for different stents, different surgical plans (e.g., different stent placements), etc.

[0171] [Plaque regression classification] Plaques can change size for a variety of reasons. For example, plaques can shrink due to densification and / or a decrease in their total mass. A decrease in the threshold resulting from (or caused by) a change in the total mass of the plaque can be classified as true regression. Other changes in the plaque can be classified as pseudoregression. Distinguishing between true and pseudoregression is difficult. For example, when a human determines whether a plaque is truly regressing or pseudoregressing, the methods used to determine plaque volume, plaque density, etc., may vary among observers. Even among the same observer, there may be variability in how plaque volume, plaque density, etc., are determined. Such intra-observer and / or inter-observer variability can lead to unreliable and / or inconsistent determinations as to whether a plaque regression is true or pseudoregression. In some embodiments, the normalization approaches described herein can be used to improve the accuracy, consistency, and / or reproducibility of regression classifications.

[0172] Disclosed herein are systems, methods, and apparatus for determining the regression and / or stabilization of plaques, such as coronary artery plaques or atherosclerotic plaques, for example, over time and / or after treatment. In some embodiments, the systems, methods, and apparatus described herein relate to plaque analysis for distinguishing between true or true regression and false regression of plaques, such as coronary artery plaques. In some embodiments, false regression of plaque may be characterized as a decrease in plaque volume due to densification (e.g., as fluffy cotton candy becomes spherical or denser), as opposed to a true decrease in plaque volume with relatively consistent material density. In some embodiments, false regression of plaque volume may be characterized as a decrease in plaque volume where the material density of the plaque has increased above a predetermined threshold amount or percentage. In some embodiments, true regression of plaque volume may be characterized as a decrease in plaque volume where the material density remains constant or relatively constant, or has not increased and / or changed above a predetermined threshold amount or percentage. In some embodiments, true regression may be characterized as a decrease in plaque volume and mass. In some embodiments, pseudo-regression may be characterized as a decrease in volume without a decrease in mass. In some embodiments, the approaches herein can be used to distinguish between pseudo-regression and true regression. For example, in some embodiments, the approaches herein can be used to classify plaque changes as true regression, pseudo-regression, stasis, or growth. Stasis may be characterized as plaque size (e.g., plaque volume) remaining constant or nearly constant, or changing below a threshold. Growth may be characterized as plaque size (e.g., plaque volume) increasing beyond a threshold (e.g., 0% or greater, 5% or greater, 10% or greater, or any other value between these values, or one or more).

[0173] In some embodiments, true plaque regression and pseudo-plaque regression can be determined based on a comparison of changes in plaque volume and plaque density, for example, based on the mathematical product of plaque volume and plaque density (e.g., plaque mass). For example, a decrease in the total mass of the plaque may indicate true regression, while a constant, near-constant, or increasing total mass of the plaque may indicate pseudo-regression or continuous growth of the plaque. In some embodiments, the mass of the material can be determined by determining the material density of the plaque based on the radiant density of the plaque and multiplying the material density by the volume of the plaque. In some embodiments, the mass of the material may not be calculated. For example, instead of converting the radiant density to a material density and multiplying the material density by the volume of the plaque, a substitute for mass can be calculated by multiplying the radiant density by the volume of the plaque.

[0174] In some cases, being able to distinguish between pseudoplaque regression and true plaque regression can be important for clinical follow-up, decision-making, and other purposes. For example, in some embodiments, both pseudoplaque regression and true plaque regression may be considered beneficial to the patient. However, clinically relevant differences may exist between the two. For instance, if most of a patient's plaque consists of calcified plaque, pseudoplaque regression may be considered to provide less improvement or reduction in the patient's risk of CAD or MACE compared to true plaque regression. However, if most of a patient's plaque consists of non-calcified plaque, low-density non-calcified plaque, or a combination of both, then plaque calcification may provide some benefit, and pseudoplaque regression may show the same or similar improvement or reduction in the patient's risk of CAD or MACE as true plaque regression. As described herein, this may be the case within certain limitations. For example, high-risk plaques consisting of fatty and fibrous tissue (e.g., low-density, non-calcified plaques) are more likely to rupture, potentially causing myocardial infarction and / or thrombosis, and can pose a greater risk to the patient than non-calcified plaques (which may have a lower likelihood of rupture). Calcified plaques, largely composed of calcium, are the least likely to rupture. Therefore, calcification of non-calcified plaques can reduce the risks associated with the plaque. However, further densification or calcification of already calcified plaque may have little to no reduction in risk to the patient.

[0175] In some embodiments, true regression of plaque volume, pseudo-regression of plaque volume, or both, may be influenced by one or more external factors or treatments, such as age, sex, exercise (exercise intensity, exercise duration, exercise frequency), diet, medication (e.g., statins, PCSK-9 inhibitors), surgical intervention (e.g., stent placement), or any combination thereof.

[0176] In some embodiments, the systems, methods, and apparatus described herein may be configured to analyze medical images to determine and / or distinguish between true plaque regression, false plaque regression, etc. In some embodiments, the systems, methods, and apparatus described herein may be configured to determine and / or distinguish between true plaque regression and / or false plaque regression based on at least one of the following: changes in total plaque volume and changes in plaque density or radiation density distribution over time or between two or more time points. In some embodiments, the systems, methods, and apparatus described herein may be configured to determine and / or distinguish between true plaque regression and / or false plaque regression based at least partially on one or more of the following: changes in total plaque volume, area, or length; changes in low-density uncalcified plaque volume, area, or length; changes in uncalcified plaque volume, area, or length; changes in calcified plaque volume; changes in material density; changes in radiation density; or changes in the Hounsfield unit density distribution of one or more of the aforementioned types of plaque over time or between two or more time points.

[0177] In some embodiments, the systems, methods, and apparatus described herein may be configured to utilize artificial intelligence (AI) and / or machine learning (ML) algorithms (also referred to herein as models) trained on datasets comprising known distinctions between true regression and pseudo-regression in plaque volume. For example, in some embodiments, a set of images showing known true and pseudo-regression can be labeled as true or pseudo-regression and used to train a machine learning model that classifies regressions as true or pseudo-regression. In some embodiments, specific features can be identified to determine the plaque regression type. In some embodiments, the user may not identify specific features, but instead can provide a relatively large number of features to the machine learning model.

[0178] [Guidance on local and systemic treatment] Furthermore, this specification describes systems, methods, and apparatus for inducing local treatment, systemic treatment, or both using image-based plaque analysis. In some embodiments, the systems, methods, and apparatus described herein relate to image-based analysis of one or more regions of plaque, such as coronary artery plaque, based on one or more distance, volume, density, radiation density, shape, morphology, degree of embedding, axial measurement, or any combination thereof. For example, in some embodiments, the systems, methods, and apparatus described herein relate to using one or more analyses of plaque to determine or induce a decision for local treatment, systemic treatment, or both of the plaque to a patient. In some embodiments, the systems, methods, and apparatus described herein are configured to utilize AI / ML, as will be described in more detail herein.

[0179] In some embodiments, the systems, methods, and apparatus described herein may be configured to determine and / or guide decisions regarding local plaque treatment, systemic plaque treatment, or both for a particular patient, at least in part, based on image-based analysis of plaque (e.g., coronary artery plaque) from medical images (e.g., CT images). In some embodiments, the systems, methods, and apparatus described herein can be used to determine patient-specific treatment or therapy. In some embodiments, treatment or therapy may include local treatment, systemic treatment, or both. For example, in some embodiments, local treatment may include percutaneous stent placement, percutaneous coronary intervention (PCI), coronary artery bypass grafting (CABG), and / or one or more of these to target treatment in one or more specific areas of plaque. In some embodiments, systemic treatment may include medical treatment, drug therapy, dietary therapy, physical exercise, and / or one or more of the same. Systemic treatment may target all plaque as a whole. In some embodiments, the systems, methods, and apparatus described herein may be configured to determine, based on image-based plaque analysis, whether a combination of topical and systemic treatment, topical treatment alone, or systemic treatment alone is recommended for a particular patient. In some embodiments, additional information such as the patient's weight, activity level, and age may be taken into consideration. For example, if a patient is already exercising regularly, recommending more exercise may be less likely to benefit that patient. As another example, a patient with mobility issues may not be able to make certain changes to their exercise routine.

[0180] In some embodiments, the systems, methods, and devices described herein may be configured to recommend specific local treatments, systemic treatments, or both to a patient in order to improve the prognosis of coronary artery disease (CAD), coronary plaque, ischemia, atherosclerosis, or any combination thereof. In some embodiments, the systems, methods, and devices described herein may be configured to recommend one or more specific local treatments, systemic treatments, or any combination thereof to a patient in order to improve hard outcomes such as death, myocardial infarction, or other adverse outcomes. In some embodiments, the systems, methods, and devices described herein may be configured to output multiple treatment options to a user (e.g., a physician) in order to provide guidance to the physician in determining a treatment plan.

[0181] In some embodiments, the systems, methods, and apparatus described herein may be configured to use a generated lesion-level risk score to guide or recommend medical decision-making in a manner that provides therapeutic benefits through local mechanical treatment by percutaneous coronary intervention (PCI) and / or with minimal or otherwise appropriate levels of systemic treatment. For example, since the lesion treated by PCI may be the only lesion, systemic treatment may not be guaranteed. In some cases, even if the current lesion is treated with local treatment, systemic treatment may be included in the treatment plan, for example, to reduce the risk of further lesion development.

[0182] In some cases and patients, a large total plaque volume at the patient level may not indicate the presence of high-risk plaques. In such patients, patient-level and lesion-level risk scores may not identify those who would benefit from local treatment with PCI. Thus, in some embodiments, the systems, apparatus, and methods described herein can be configured to recommend managing such patients solely with systemic medical treatment.

[0183] In some cases and patients, the total plaque volume at the patient level may be high, and high-risk local plaques may also be present. In these patients, patient-level and lesion-level risk scores can distinguish between lesions that may benefit from local treatment with PCI and lesions that are not eligible for PCI but may benefit from systemic treatment. Thus, in some embodiments, the systems, apparatus, and methods described herein may be configured to recommend that both local PCI and systemic medical treatment may benefit the patient.

[0184] In some embodiments, the systems, apparatus, and methods described herein can utilize image-based plaque and / or atherosclerosis analysis to determine patient-specific needs, including systemic treatment (e.g., medication, dietary changes, exercise, etc.), local treatment (e.g., PCI or CABG), both, or neither.

[0185] In some embodiments, the risk stratification approach described herein can be used to generate risk stratifications for various systemic treatments, topical treatments, or both. For example, in some embodiments, the risk stratification approach described herein can be used to determine the prognosis for various interventions for a particular patient and can assist the intervener in determining an appropriate treatment plan.

[0186] [Reconstruction of the coronary artery tree] Furthermore, this specification discloses systems, methods, and apparatus for reconstructing the coronary artery tree. Accurately constructing the coronary artery tree can be important. For example, accurately determining the structure of a patient's coronary artery tree can be important for diagnosing coronary artery disease, evaluating the progression of coronary artery disease, and determining treatment plans for coronary artery disease.

[0187] The coronary artery tree refers to the network of blood vessels that supply oxygen and nutrients to the heart muscle. Examples of coronary artery trees include the left coronary artery (LCA), the right coronary artery (RCA), and the left circumflex artery (LCx). The coronary arteries can branch into smaller vessels and capillaries, forming a complex network.

[0188] Impaired blood flow in the coronary arteries can lead to coronary artery disease (CAD) or coronary heart disease (CHD). The coronary arteries can become narrowed or blocked by the accumulation of fatty deposits, cholesterol, plaque, or a combination thereof.

[0189] There are various techniques for imaging the coronary arteries, such as computed tomography angiography (CCTA), coronary angiography, and magnetic resonance angiography (MRA). While these imaging techniques are useful for constructing a coronary artery tree, there are many challenges in constructing a coronary artery tree. Blood vessels are small and can have various winding and complex anatomical structures. The heart is constantly moving due to its pumping action. The pumping action of the heart can introduce blurring and other artifacts into the image. In some cases, echocardiography (ECG) can be used to trigger image acquisition so that images are acquired during specific phases of the heartbeat (e.g., during contraction (systole), during relaxation (diastole), or between phases). In some embodiments, images can be acquired at specific points in the heartbeat, such as at maximal contraction or maximal relaxation.

[0190] Blood vessels can be located in close proximity to other tissues and structures, such as ventricles, valves, and surrounding tissues. Distinguishing blood vessels from neighboring structures is difficult. While contrast agents can sometimes be used to improve visibility, obtaining sufficient contrast is difficult, especially in the presence of calcified plaque. Image acquisition requires the patient to remain still for a considerable amount of time (e.g., up to approximately 30 minutes). During acquisition, the patient may move, breathe, or change position, resulting in blurring and other image artifacts.

[0191] Problems with images (e.g., blurring, noise, poor contrast, or other issues) can make it difficult or impossible to extract blood vessels, resulting in the rejection of the images. If images are rejected, patients may have to return for additional imaging, which can be frustrating, cause delays, incur significant costs, and / or occupy potentially limited resources, as the hospital or other medical facility may only have a limited number of devices capable of collecting images. In some cases, if images of sufficient quality are not obtained, computational analysis of the patient's images may simply not be performed, in which case physicians may make decisions without such information, potentially worsening the patient's outcome.

[0192] Images acquired during coronary imaging procedures may, in some cases, be divided into multiple series that may correspond to different cardiac phases, such as systole or diastole. A certain vessel may be adequately imaged in the first series (e.g., corresponding to the first cardiac phase), while a different vessel may be adequately imaged in a second, third, fourth, and so on series. While appropriate imaging information may exist for each vessel, it can be difficult to determine which series to use for which vessel when the imaging information spans different series.

[0193] Even if blood vessels can be identified in different series and high-quality images can be extracted, the usefulness of extracting individual vessels or groups of vessels may be limited because it does not provide a complete picture of the coronary artery tree. Conventional approaches have sometimes failed to create a complete coronary artery tree by combining images from different series, or have resulted in anatomically inaccurate coronary artery trees.

[0194] Furthermore, while it is important to co-register different series to create an accurate coronary artery tree, this can be difficult because blood vessels tend to shift with the heartbeat and patient movement. In some embodiments, a common reference point can be used to register multiple series.

[0195] In some embodiments, a machine learning model or multiple machine learning models can be trained to extract vessels from CCTA images or other coronary artery images. The extracted vessels can then be subjected to further processing for use in reconstructing a coronary artery tree. As described herein, images can be divided into multiple series. In some embodiments, vessels can be extracted from each series of multiple series.

[0196] After extracting the blood vessels, it is important to identify which series provides the highest quality image for each vessel. For example, the left anterior descending (LAD) artery and the left circumflex coronary artery can be imaged with higher quality in the first series (e.g., the series corresponding to diastole), while other vessels such as the left LCA can be imaged with higher quality in a different series (e.g., systole).

[0197] In some embodiments, each blood vessel can be labeled. In some embodiments, the labels can correspond to specific anatomical features, but this is not mandatory. For example, in some embodiments, blood vessels can be labeled as LAD, LCA, etc. In some embodiments, blood vessels can be labeled with numbers, letters, or other labels (e.g., blood vessel 1, blood vessel 2, etc.). Labeling blood vessels is important because it allows for comparison of specific blood vessels or groups of blood vessels across multiple image sequences.

[0198] Selecting the optimal sequence for each vessel can be challenging and may involve considerations such as motion blur, contrast, and image noise. For example, in some embodiments, the best sequence for each vessel can be selected, at least partially, by minimizing motion blur, maximizing contrast, minimizing image noise, and / or based on these. In some embodiments, a machine learning model or multiple machine learning models can be used to select the optimal sequence for each vessel. In some embodiments, the first step in selecting a series for each vessel may include straightening each vessel. The straightened vessel images can then be input into a machine learning model trained to identify the best sequence from multiple sequences. In some embodiments, the vessels may not be straightened before being input into the machine learning model, and the machine learning model can be trained to identify the best sequence from multiple sequences without straightening the vessels.

[0199] In some embodiments, the aorta can be segmented from images in one or more image sequences. In some embodiments, the aorta may be large compared to the vessels of the coronary artery tree, so its extraction may be relatively easy. In some cases, a contrast agent may be used during the image acquisition procedure, and as a result, the aorta may be easily visualized in the images.

[0200] While the above operations can be used to extract blood vessels and identify the best series of each vessel, it may be important to combine or register the images of each vessel to reconstruct a complete coronary artery tree. In some embodiments, the aorta can be used as a reference for combining images to generate a reconstructed complete (or nearly complete) coronary artery tree. In some embodiments, the aorta can be used as a reference (one or more) for reconstructing the coronary artery tree. In some embodiments, images from different series can be combined at least partially based on one or more reference points or landmarks, such as one or more branching landmarks. The reconstructed coronary artery tree can be used for a variety of purposes.

[0201] In some embodiments, the systems and methods described herein can be used for plaque visualization. For example, plaque can be superimposed on a complete or nearly complete coronary artery tree. In some embodiments, the systems and methods described herein can enable more accurate fractionated flow reserve CT (FFR-CT) analysis, which can be used to determine or estimate the effects of coronary artery stenosis. In some embodiments, the systems and methods described herein can be used to assess or visualize critical myocardial volume (MMAR), which may require an accurate and complete or nearly complete coronary artery tree.

[0202] [Thresholds for variable plaque classification] Plaque calcification is an important indicator of disease and helps identify the risk of cardiac adverse events, for example. Correctly identifying calcified plaque can have significant implications for medical interventions, such as determining whether a stent should be placed in the blood vessel, whether procedures such as atherectomy or angioplasty should be performed, and what type of stent to use. In some embodiments, plaque classification can indicate whether systemic intervention, local intervention, or both should be pursued.

[0203] While plaque classification is important, determining whether a plaque is calcified and to what extent is difficult. Invasive procedures such as optical coherence tomography (OCT) can provide relatively accurate and / or reproducible determination of the presence and size of calcified plaques, but such invasive procedures are time-consuming and carry risks of procedural complications. Furthermore, because OCT generally has limited tissue penetration, it may not be able to fully capture the characteristics of the plaque.

[0204] Computed tomography angiography (CCTA) offers high spatial and temporal resolution, providing detailed information about plaque morphology. However, CCTA has limitations in its ability to distinguish between calcified and non-calcified plaques, and the apparent size of the plaque can vary depending on image acquisition parameters, attenuation within the patient's body, and attenuation in the surrounding area (e.g., within the lumen). In some embodiments, plaques can be characterized based on their Hounsfield intensity (e.g., classified as calcified or non-calcified), but appropriate characterization can vary depending on attenuation, imaging parameters, etc. Peak kilovolts (kVp) can significantly affect the apparent size of calcified plaques.

[0205] Therefore, while CCTA can provide valuable insights, its usefulness in classifying plaque as calcified or non-calcified, or in determining the degree of calcification, may be limited. This specification describes approaches that can be used to mitigate or overcome such limitations, and that by making CCTA images usable to obtain valuable insights about plaque, the need for invasive procedures can be reduced or eliminated.

[0206] In some embodiments, the approaches described herein can be used to normalize CCTA images and provide reliable, reproducible insights into plaque calcification. In some embodiments, the approaches described herein can improve the accuracy of identifying and / or classifying calcified plaques.

[0207] In CCTA, the boundary between non-calcified and calcified plaques (or between plaques with different calcification levels) can be defined by radiation intensity. In some embodiments, the boundary can be defined in terms of Hounsfield units (HU). In some embodiments, a suitable boundary can be defined in relation to peak kilovolts and luminal HU values. For example, it has been observed that luminal contrast affects the apparent size of plaque. Therefore, the boundary between plaque types (e.g., between calcified and non-calcified plaques) may depend on the observed luminal contrast. Changes in peak kilovolts (kVp) can significantly affect the X-ray spectrum produced by the X-ray source, and consequently, can significantly affect the CCTA image.

[0208] Many factors can influence the appropriate threshold for identifying calcified plaques and distinguishing them from uncalcified plaques. For example, specific X-ray sources, X-ray detectors, collimators, and geometry (e.g., distance from source to detector) can each affect the final CCTA image. In some embodiments, the boundary between calcified and uncalcified plaques can be determined for specific X-ray detectors, sources, and geometric configurations. However, such an approach may be undesirable because it is difficult to predict all possible combinations of equipment or equipment components that may be used to acquire CCTA images. Therefore, in some embodiments, it may be desirable to reduce or eliminate the need to consider specific types of equipment, configurations, etc. As described herein, in some embodiments, the plaque calcification threshold can be defined by kVp and lumen contrast. Using these two parameters, the system can be configured to determine an appropriate boundary between calcified and uncalcified plaques, thereby providing a relationship between kVp, lumen contrast, and plaque calcification threshold that is independent of specific equipment characteristics and parameters.

[0209] Luminous contrast can be an attractive parameter to use when determining plaque classification thresholds, because, for example, the lumen is an empty cavity and is therefore expected to be shown fairly consistently when imaged.

[0210] In some embodiments, OCT images can be used as a baseline or ground truth for the true range of calcified plaque. However, it will be understood that OCT may not provide a fully accurate measurement of calcified plaque. Nevertheless, OCT can be accepted as a suitable criterion for determining the threshold between calcified and non-calcified plaque.

[0211] While OCT can be used to differentiate between calcified and non-calcified plaques, it can struggle to differentiate between low-attenuation plaques (LAPs) and fibrous, fatty plaques. In some embodiments, near-infrared spectroscopy (NIRS) can be used to distinguish between LAPs and fibrous, fatty plaques. The approaches described herein can be readily adapted to other imaging modalities, thereby enabling the use of CCTA in a wide range of applications.

[0212] The approach described herein can be used to more accurately and / or reproducibly determine the presence, size, etc., of calcified plaques using non-invasive CCTA imaging. In some embodiments, the approach described herein can be used to characterize the volume, area, and / or length of calcified plaques. For example, in some embodiments, the approach described herein can be used to facilitate a more accurate and / or reproducible determination of plaque thickness (e.g., thickness extending from the vessel wall into the lumen).

[0213] Much of the discussion herein concerns the differentiation between calcified and non-calcified plaques (e.g., the identification of plaque calcification thresholds for differentiating between calcified and non-calcified plaques), but the approaches herein are not so limited. In some embodiments, the approaches herein may be used to distinguish between low-density calcified plaques, medium-density calcified plaques, high-density calcified plaques, and so on.

[0214] Generally, thresholds for distinguishing plaque types (e.g., non-calcified, low-density calcified, medium-density calcified, and high-density calcified) are expected to shift in the same direction as kVp and luminal density change. However, they do not necessarily shift by the same amount. In some embodiments, the relationships can be determined using the techniques described herein, which can determine multiple thresholds. For example, when adjusting thresholds to distinguish between calcified and non-calcified plaques, other thresholds (e.g., between low-density calcified and medium-density calcified plaques, or between medium-density calcified and high-density calcified plaques), using known relationships that may be linear, polynomial, exponential, or logarithmic, can be shifted by an appropriate amount.

[0215] In some embodiments, the threshold can be set at a global level (e.g., for the entire image). In some embodiments, the threshold can be set at a local level, which helps to account for local effects such as blooming that may be present around the lumen. Generally, blooming affects the lumen concentrically or nearly concentrically, and the blooming may fall with a known or determinable gradient.

[0216] In some embodiments, the system may consist of one or more default thresholds for distinguishing plaque types. In some embodiments, the system may include a graphical interface, configuration files, etc., that allows a user (e.g., a physician) to customize one or more thresholds, thereby making it easier for the physician to utilize their experience and expertise when analyzing plaque.

[0217] In some cases, when determining a threshold, there may be one or more outlier thresholds (e.g., thresholds that are significantly higher or lower than those shown by other images with similar imaging conditions). In some embodiments, one or more outlier thresholds can be discarded when determining a plaque classification threshold.

[0218] [Identification of thin capsular fibrous hemangiomas] This disclosure provides an approach for detecting capsular fibromas (TCFAs) using coronary computed tomography angiography (CCTA) imaging. The approach described herein may enable non-invasive detection of TCFAs.

[0219] Fibrous atheromas are a type of plaque that can pose a significant risk of adverse health events. Some fibrous atheromas are more likely to cause adverse health events than others. Fibrous atheromas are characterized by the presence of a fibrous cap that forms the boundary between the core and lumen of the fibrous atheroma. The thickness of the cap can be an indicator of the likelihood of the fibrous atheroma erupting. When a fibrous atheroma ruptures, the thrombosed core of the fibrous atheroma is released into the lumen, causing thrombosis. Studies have shown that ruptured plaques are characterized by the presence of a thin fibrous cap that is 65 μm or less in thickness or approximately 65 μm. Fibrous hemangiomas with such thin caps are classified as thin-capsulated fibrous hemangiomas (TCFAs).

[0220] Accurately imaging thin capsules is difficult. For example, the axial resolution of intravascular ultrasound virtual histology (IVUS-VH) is approximately 200 micrometers, which limits its ability to identify TCFAs. CCTA also lacks the ability to reliably image thin capsules.

[0221] According to some embodiments described herein, TCFAs can be identified without necessarily imaging the capsule itself and / or measuring the capsule thickness with high precision. For example, the presence of low-attenuation, non-calcified plaque near the lumen can indicate the presence of TCFAs. However, identifying low-attenuation plaques can be challenging in CCTA. Invasive procedures such as optical coherence tomography are more reliable but complex and carry patient risks.

[0222] Determining plaque density, calcification, or both can be achieved using CCTA imaging according to several embodiments described herein. The apparent size, density, etc., of the plaque may be influenced by the contrast of the nearby lumen, the peak kilovolts of the X-ray source used to acquire the CCTA image, etc. According to some embodiments, the CCTA image can be normalized based on lumen contrast (which can be measured in Haunsfield units (HU)) and kVp, for example, with respect to the classification of variable plaques as described above. In some embodiments, low-attenuation plaques (LAPs) can be distinguished from other non-calcified and / or calcified plaques.

[0223] According to some embodiments, TCFA can be determined by the presence of LAP at or near the plaque boundary, for example, at the interface between the plaque and the lumen.

[0224] Plaque types can be classified in CCTA images based on attenuation measured by HU. Generally, lower HU values ​​indicate less calcification and / or lower plaque density, as well as other differences in the plaque. While CCTA alone can provide relative information about plaque (e.g., more or less calcification compared to other plaques), it is important to measure plaque characteristics (e.g., calcification) more absolutely in order to accurately classify plaques and make treatment decisions based on reliable and accurate information.

[0225] In some cases, OCT images may not be able to resolve thin coatings, but they may be useful for distinguishing LAP from other non-calcified plaques. In some embodiments, the threshold between LAP and other non-calcified plaques in CCTA images can be determined, for example, based on the analysis of OCT images.

[0226] In some embodiments, a set of CCTA images and a set of OCT images can be co-registered, and a threshold cutoff (e.g., HU units) between the LAP and other non-calcified plaques can be adjusted relative to the CCTA images so that the CCTA images reflect the extent of the LAP determined using OCT. In some embodiments, only cross-sectional images can be used. Cross-sectional images can show the approach of the LAP to the boundary between the plaque and the lumen. In some embodiments, three-dimensional images can be used, which can provide additional and / or different insights into the extent of the LAP.

[0227] In some embodiments, it may not be necessary to know the exact size and shape of the LAP to identify TCFA. For example, it may be sufficient to determine that the LAP extends to or near the plaque surface. In some embodiments, a set of OCT images can be grouped, with one group containing images where TCFA is present and the other group containing images where TCFA is not present. In some embodiments, thresholds for distinguishing LAP from other non-calcified plaque can be adjusted so that CCTA images can be reliably classified as showing or not showing TCFA, as determined, for example, by comparison with OCT images. In some embodiments, the system can be configured so that the user (e.g., a physician) can more easily utilize their expertise and / or experience when searching for the presence of TCFA.

[0228] In some embodiments, the apparent thickness of the cap can be determined from CCTA images. As described herein, such thickness may not be accurate, but accurate thickness measurements or precise information regarding the size and shape of the LAP may not be necessary to detect TCFA. Rather, it may be sufficient to have a clear difference in the apparent cap thickness between CCTA images with and without TCFA, and / or that the location of the LAP (e.g., relative to the boundary between the plaque and the lumen) is determined accurately enough to distinguish between TCFA and non-TCFA.

[0229] In some embodiments, the system can determine the probability of TCFA. In some embodiments, the system can determine that TCFA is possible if the probability is above a threshold. In some embodiments, the system can determine that TCFA is unlikely if the probability is below a threshold. In some embodiments, the system can be configured to output an indication that TCFA is likely, an indication that TCFA is unlikely, and / or the probability of TCFA.

[0230] [Stent selection and surgical planning] Furthermore, this specification discloses systems, methods, and apparatus for analyzing lesions. Such analyses can be used for stent selection, surgical planning (e.g., manual and / or robotic surgical planning), and so on.

[0231] Coronary artery disease is a serious problem. Stent placement is beneficial in reducing the risk of adverse events, but it has significant challenges. For example, it is important to select a stent based on factors such as plaque length, normal diameter of the stenting vessel, plaque density, and plaque location. There are many types of stents available, each with different mechanical properties. It is important to select a stent with mechanical properties that are suited to the patient's specific needs. For example, it is important to consider how the stent behaves when it is flexed, under tension, compressed, or bent.

[0232] It is important to avoid using stents that are too short to completely cover the plaque. Uncovered plaque can lead to adverse outcomes such as in-stent restenosis (where the blood vessel begins to narrow again at or near the site where the stent is placed) and / or other complications.

[0233] In conventional approaches, many decisions are typically made during the intervention rather than being planned in advance. For example, interventionists may use techniques such as intravascular ultrasound (IVUS) or fluoroscopy during the procedure to determine the placement of the stent. Insufficient pre-procedure information can lead to suboptimal treatment, such as the use of stents that are too short, too long, too narrow, or too wide, wasted time during the procedure, and postoperative complications (including major cardiac adverse events).

[0234] Poor pre-procedure planning means that interventional stents do not know in advance what is needed to perform the procedure, which can lead to wasted medical supplies, wasted preparation time, and an inability to optimize patient outcomes. For example, surgical support staff can select several stents in advance, allowing the interventional stent to choose the best available stent during the procedure. Interventional stents may not have a stent of the required length and / or diameter, so they may choose the closest one. If the stent is too short, the patient is at risk of restenosis. Complications can also occur if the stent is too long. For example, the risk of thrombosis is associated with stent length, as is the risk of myocardial infarction.

[0235] The approach described herein can provide more accurate, thorough, and / or complete planning for stent placement procedures. Using the approach described herein, the length, diameter, position, etc. of the stent can be determined before the procedure is performed. In some embodiments, the positioning of the stent can be determined in advance. As described herein, positioning can be performed without using an absolute reference frame. Rather, the positioning of the stent can be determined based on the relative distance from one or more reference points. Reference points may include, for example, Ostium, the left main trifunction, the left main bifunction, or other reference points.

[0236] In some embodiments, the approaches described herein can be used, for example, in surgical training to train new or otherwise inexperienced surgeons to perform stent placement procedures. In some embodiments, the approaches described herein can be used in robotic surgery. For example, the approaches described herein can be used to generate commands that instruct a surgical robot to perform one or more movements to a desired position for stent placement.

[0237] The approach described herein can offer many advantages, including reduced medical material waste, shorter procedure times, and / or improved patient outcomes. For example, the approach described herein can reduce the risk of restenosis, thrombosis, myocardial infarction, and / or other adverse outcomes. Costs can be reduced because treatment facilities can order stents only for planned procedures where the required stents are known in advance.

[0238] Traditionally, methods such as fractional flow reserve (FFR) have been used to analyze stenosis. FFR measures the pressure difference across the stenosis to determine the likelihood that the stenosis is significantly reducing oxygen supply. FFR is an invasive procedure in which a catheter is inserted into the patient and guided to the patient's heart. Various other techniques can be used additionally or alternatively to obtain information for diagnosing and treating the patient. For example, invasive angiography can be used to determine stenosis, ultrasound to determine plaque length and volume, near-infrared spectroscopy to identify lipid-rich plaques, and optical coherence tomography to determine the minimum lumen diameter.

[0239] In some embodiments, the approaches described herein can be used to non-invasively obtain information similar to that previously only obtainable through invasive procedures. The techniques described herein can be used for ischemia identification, lesion identification (e.g., lesions of the left main artery, lesions of the left anterior descending artery, etc.). In some embodiments, the approaches described herein can be used to distinguish between plaque types (e.g., between calcified and non-calcified plaques), determine the required stent length, and / or determine whether multiple stents are needed. In some embodiments, data from multiple data acquisition modalities can be included in the same user interface, thereby facilitating data interpretation.

[0240] In some embodiments, surgical decisions can be made by analyzing CCTA images. However, the information is often scattered across multiple locations, making planning difficult. In some embodiments, it is useful for the user to display multiple types of information on a single screen to determine the stent placement procedure, stent length, stent design, and whether or not to intervene. In some embodiments, the system can be configured to identify the possibility of ischemia. In some cases, the interventionist can decide to perform stent placement only on plaques that are likely to be ischemic. By using the approach described herein, surgical procedures can be planned in advance, and the appropriate size, material, and shape of the stent can be selected, thus providing clinical and economic benefits. For example, the patient may experience fewer complications, and the facility can stock only the necessary stents. In some cases, the interventionist may want to calcify soft plaque. Using the technique described herein, the interventionist can identify soft plaque and select a drug-eluting stent containing, for example, an agent that helps calcify the plaque.

[0241] According to some embodiments, a user interface can be provided that illustrates blood vessels and labels them to indicate conditions such as stenosis, potential ischemia, and chronic total occlusion. In some embodiments, a user interface can be provided that shows an elongated diagram of a blood vessel and identifies lesions (e.g., plaques) within the vessel. In some embodiments, the user can click on the lesion or otherwise select it to view information about the lesion, such as maximum diameter stenosis, plaque length, minimum lumen diameter, normal lumen diameter, plaque volume, and calcification.

[0242] As described herein, in some embodiments, the system may be configured to generate a surgical plan. For example, in some embodiments, a user can select one or more plaques for treatment and generate a treatment plan. The treatment plan may include, for example, the location of the plaque (e.g., distance from a reference point such as osstium), the length of the stent, and / or the diameter of the stent.

[0243] In some embodiments, the system may be configured to generate a robotic surgical plan. The robotic surgical plan may include, for example, instructions for the movement of the surgical robot. For example, the surgical robot may be instructed to follow a path along a blood vessel until it reaches a specific distance from a reference point (e.g., the periosteum).

[0244] [Examples] Figure 7 is a block diagram illustrating exemplary processes for distinguishing between true plaque regression and pseudo-plaque regression according to several embodiments. The processes shown in Figure 7 can be performed on a computer system. In some embodiments, the operations shown in Figure 7 may be performed in a different order than shown, additional operations may exist, and / or one or more illustrated operations may be omitted or combined with other operations.

[0245] In operation 1010, the computer system can receive a first medical image. In operation 1020, the computer system can receive a second medical image. The first and second medical images may include areas of plaque. In operation 1030, the computer system can identify a first area of ​​plaque in the first medical image. In operation 1040, the computer system can determine a first set of one or more characteristics of the first area of ​​plaque. These one or more characteristics may include, but are not limited to, total plaque volume, low-density non-calcified plaque volume, non-calcified plaque volume, calcified plaque volume, material density, radiation density, Hounsfield unit density of one or more types of plaque, or any combination thereof. In operation 1050, the computer system can identify a second area of ​​plaque in the second medical image. The second area of ​​plaque may be the same plaque area as the first area of ​​plaque. The second medical image may be, for example, an image of the same region of the plaque at a later point in time. In operation 1060, the computer system may determine a second set of one or more characteristics of the second region of the plaque. In operation 1070, the computer system may determine a change in at least one characteristic between the first medical image and the second medical image by comparing at least one characteristic of the first set of characteristics with at least one corresponding characteristic of the second set of characteristics. In operation 1080, the computer system may classify whether the observed plaque regression is a true regression or a pseudo-regression based on at least one characteristic comparison.

[0246] Figure 8 is a block diagram illustrating an exemplary process for determining a recommended treatment according to several embodiments. The process shown in Figure 8 can be performed using a computer system. In some embodiments, the operations shown in Figure 8 may be performed in a different order than shown, additional operations may exist, and / or one or more illustrated operations may be omitted or combined with other operations.

[0247] In operation 1110, the system can access a medical image containing representations of one or more arteries and one or more regions of plaque (e.g., one or more regions of atherosclerotic plaque). In operation 1120, the computer system can analyze the medical image to identify one or more arteries and one or more regions of plaque. In some embodiments, identifying one or more arteries may include determining anatomical markings for one or more arteries, but this is not required and may not occur in some other embodiments. In operation 1130, the computer system can determine the total plaque volume. In operation 1140, the computer system can determine the extent of high-risk plaque. In operation 1150, the system can assess the expected therapeutic effect of systemic treatment based, for example, the total plaque volume, the total high-risk plaque volume, or both. In some embodiments, the expected therapeutic effect may take into account other factors such as the patient's age, weight, activity level, and previous interventions. In operation 1160, the system can assess the expected therapeutic effect of local treatment based, for example, the total plaque volume, the extent of high-risk plaque, or both. In some embodiments, the computer system may take into account other factors such as the patient's age, weight, activity level, and previous interventions. In operation 1170, the computer system may determine a recommended treatment or multiple recommended treatments based on the expected therapeutic benefits of systemic treatment and local treatment. In some embodiments, the computer system may be able to present the recommended treatment or multiple recommended treatments to the physician or other user. In some embodiments, systemic treatment may be preferred over local treatments such as stent placement because the likelihood of complications from medication, dietary changes, or exercise changes may be lower. In some embodiments, local treatment may be preferred when systemic treatment is unlikely to be effective and / or when systemic treatment may be difficult, such as when drug interactions, limitations in the patient's physical activity, or when it has been demonstrated that the patient does not adhere to a diet or exercise plan.

[0248] Figure 9 is a flowchart illustrating exemplary image normalization processes according to several embodiments. The exemplary processes in Figure 9 can be executed on a computer system.

[0249] In operation 1210, the system can access the patient's CT image. In some embodiments, the CT image may include representations of one or more arteries. In some embodiments, one or more arteries may include one or more regions containing atherosclerotic plaques. In operation 1220, the system can access one or more image acquisition parameters used to acquire the CT image. In some embodiments, the image acquisition parameters may include, for example, but are not limited to, the helical CT scheme, the type of CT detector, the type of CT based on the number of photon energy spectra, the X-ray generator current (mA), peak kilovolts (kVp), image noise, image signal, signal-to-noise ratio, contrast-to-noise ratio, and / or contrast enhancement. In operation 1230, the system can normalize the CT image by applying an image processing algorithm to the CT image. In some embodiments, normalization may not utilize a physical calibration device. In some embodiments, the image processing algorithm can be derived by analyzing multiple test CT images obtained from the same subject and one or more image acquisition parameters used to obtain the multiple test CT images. In some embodiments, the test CT image may include one or more regions containing atherosclerotic plaques. In some images, the image processing algorithm can be derived by analyzing the simulated CT image.

[0250] In some embodiments, normalized CT images can be analyzed to generate one or more plaque parameters and / or one or more vascular parameters. The one or more plaque parameters may include, for example, total plaque volume, non-calcified plaque volume, and / or calcified plaque volume. In some embodiments, the one or more vascular parameters may include one or more lumen measurements. In some embodiments, the one or more plaque parameters and / or one or more vascular parameters generated from normalized CT images can be compared to one or more corresponding plaque parameters and / or vascular parameters generated from other (e.g., previous) CT images of the patient.

[0251] Figure 10 is a flowchart illustrating an exemplary process for generating an image processing algorithm according to several embodiments (for example, the image processing algorithm described above with reference to Figure 9). The exemplary process in Figure 8 can be executed on a computer system.

[0252] In operation 1310, the system can access multiple test CT images obtained from the same subject. In some embodiments, the multiple test CT images may include representations of one or more arteries of the subject. In some embodiments, one or more arteries may include one or more regions containing atherosclerotic plaques. In operation 1320, the system can access one or more image acquisition parameters used to acquire the multiple test CT images. In operation 1330, the system can generate an image processing algorithm to normalize the patient's CT images based at least partially on the multiple test CT images. In some embodiments, physical calibration device images may not be used to generate the image processing algorithm.

[0253] In some embodiments, the image processing algorithm can be generated using regression (e.g., linear regression or nonlinear regression). In some embodiments, the image processing algorithm can be generated by training a machine learning model. For example, a test image and its image acquisition parameters (which may be real or simulated images) can be used as input to train a machine learning model (for example, the test image and image acquisition parameters can be used to generate feature vectors for the machine learning model). In some embodiments, the generated image processing algorithm can receive two sets of image acquisition parameters associated with a first image and a second image, and the generated image processing algorithm can use the received image acquisition parameters to determine a normalization parameter to apply to either the first or second image in order to normalize the first and second images with respect to each other.

[0254] Figure 11 is a block diagram illustrating an exemplary process for determining risk stratification according to several embodiments. The process shown in Figure 11 can be performed on a computer system. In operation 1410, the computer system can access the subject's medical images. In operation 1420, the computer system can analyze the medical images to identify one or more areas of stenosis in one or more coronary arteries. In operation 1430, the computer system can analyze the medical images to determine one or more areas of atherosclerosis in one or more coronary arteries. In operation 1440, the computer system can analyze the medical images to determine one or more areas of ischemia in one or more coronary arteries. In operation 1450, the system can generate a MACE risk stratification for the subject. In some embodiments, the system can provide the generated risk stratification to a user, such as a physician.

[0255] Figure 12 is a flowchart illustrating an exemplary process for reconstructing a coronary artery tree according to several embodiments. The process shown in Figure 12 can be performed on a computer system using multiple images. These images may be, for example, computed tomography (CCTA), coronary angiography, and / or magnetic resonance angiography (MRA) images.

[0256] In operation 1510, the method includes extracting coronary arteries from multiple images corresponding to multiple series. The method may include extracting coronary arteries for each series. Each series may correspond to a phase of the heartbeat. For example, images can be acquired in response to an ECG trigger so that each image series corresponds to a specific phase of the heartbeat. In embodiments, one or more machine learning models can be used to extract coronary arteries. In some embodiments, one or more machine learning models may include a convolutional neural network.

[0257] In operation 1520, the method includes labeling the blood vessels extracted in each series. In some embodiments, the labeling may correspond to anatomical features. In some embodiments, the labeling may not correspond to anatomical features. The labeling may be the same for the same blood vessel across multiple series.

[0258] In operation 1530, the method includes segmenting the aorta. For example, the patient's aorta can be viewed in each of several series. The aorta or a portion thereof can be used as a reference when reconstructing the complete coronary artery tree.

[0259] In operation 1540, the method includes ranking each series of multiple series for each labeled vessel. The ranking of the series may be performed using a machine learning model or multiple machine learning models. These machine learning models may include convolutional neural networks, deep neural networks, etc. The highest-ranking image series may be the images best suited for reconstructing the coronary artery tree. The highest-ranking image series may, for example, have higher contrast, less image noise, and / or less blue than other series.

[0260] In some embodiments, the method may include straightening of each vessel before ranking the series. For example, images of vessels may undergo geometric transformations to make the vessels appear straight.

[0261] In operation 1550, the method includes registering selected sequences for each vessel in order to generate a reconstructed coronary artery tree. For example, the aorta or one or more portions thereof can be used as reference points for reconstructing the coronary artery tree. For example, when determining the relative positioning of the left and right major coronary arteries, the left and right major coronary arteries immediately above the coronary ostiae, and the coronary ostiae themselves can be used. In some embodiments, the overlap ratio can be used to determine the quality of the reconstructed coronary artery tree.

[0262] Figure 13 is an explanatory diagram of the reconstructed coronary artery tree and the series used to generate the reconstructed coronary artery tree. In Figure 13, there are four series, and the vessels are divided into three vessels. In some embodiments, there may be more or fewer series and / or more or fewer vessels. In Figure 13, areas of poor image quality are indicated by ellipses. In Figure 13, Series 1 is used for the first vessel, Series 2 is used for the second vessel, and Series 4 is used for the third vessel. Series 3 is not used for any vessels in the reconstructed coronary artery tree because each vessel has areas of poor image quality. In Figure 8, there is a one-to-one correspondence between series and vessels, but it will be understood that one series may be the best series for multiple vessels. For example, in the illustrated example, Series 1 and Series 2 are drawn with similar image quality for the first vessel, and Series 2 may be the best series for both the first and second vessels.

[0263] Figure 14 is a flowchart illustrating an exemplary process for adjusting the plaque classification threshold in CCTA images according to several embodiments. This process can be performed by a computer system. In the following discussion, we assume that the threshold is set in HU. However, it will be understood that any suitable scale describing attenuation in CCTA images can be used.

[0264] In operation 1705, the system can receive a set of OCT images and a set of CCTA images. Each OCT image can form an image pair corresponding to the CCTA image. In operation 1710, the system can co-register each set of OCT and CCTA images. In operation 1715, the system can determine the size of the calcified plaque from the OCT image. In some embodiments, the calcified plaque size may have been previously determined, in which case operation 1715 can be skipped or modified to include receiving the previously determined calcified plaque size. In operation 1720, the system can determine the size of the calcified plaque from the received CCTA image. In operation 1725, the system can adjust the CCTA plaque classification threshold so that the calcified plaque size determined from the CCTA image more closely matches the calcified plaque size determined from the OCT image.

[0265] Although not shown in FIG. 14, as described herein, it will be understood that the threshold can vary depending on the kVp and lumen contrast used to acquire the image. Thus, in some embodiments, the process depicted in FIG. 14 can be performed for multiple combinations of kVp and lumen contrast to generate multiple plaque classification thresholds corresponding to different combinations of kVp and lumen contrast. Typically, the kVp value is limited to a few values such as 80 kV, 100 kV, or 120 kV. However, the lumen contrast can vary from image to image. Thus, in some embodiments, the lumen contrast can be binned into groups having similar contrast. For example, the bins can include about 20 HU, about 25 HU, about 50 HU, about 100 HU, or any value between these values, or ranges above or below these values.

[0266] In some embodiments, the lumen contrast value may not be binned. For example, in some embodiments, the determined threshold can be used to calculate an equation that outputs a plaque classification threshold and takes lumen contrast and kVp as inputs. In some embodiments, the equation can be a linear equation, a polynomial equation, a logarithmic equation, an exponential equation, etc.

[0267] Figure 15 is a drawing schematically showing the influence of plaque calcification thresholds according to some embodiments. In Figure 15, a first CCTA image 1802 having a first plaque calcification threshold shows a first calcified plaque region 1808, and the second CCTA image 1804 can be the same image as the first CCTA image 1802 but has a second different plaque calcification threshold. The second CCTA image 1804 can show a second calcified plaque region 1810. The first plaque calcification threshold can be lower than the second plaque calcification threshold, whereby the first calcified plaque region 1808 appears larger than the second calcified plaque region 1810. The OCT image 1806 can be taken as a "ground truth" image. As discussed herein, the OCT image may not represent the true size of the calcified plaque region, but in some cases can be considered a good approximation. In the OCT image 1806, a calcified plaque region 1812 is illustrated. As shown in Figure 15, the size of the first calcified plaque region 1808 is close to the size of the calcified plaque region 1812. This can be because the first plaque calcification threshold more accurately represents the threshold between calcified and non-calcified plaques than the second threshold, and the plaques identified as calcified in the OCT image 1806 are identified as non-calcified in the second CCTA image 1804.

[0268] Figure 16 is a schematic diagram illustrating various possible thresholds for plaque according to several embodiments. As shown in Figure 16, typically, more calcified plaques exhibit higher contrast (e.g., higher HU values). In some embodiments, plaques are divided into two categories: calcified plaques and non-calcified plaques. However, it will be understood that there is no clear cutoff between calcification and non-calcification, and the level of calcification can lie on a spectrum. In some embodiments, calcified plaques can be subdivided, for example, into low-density calcified plaques, medium-density calcified plaques, and high-density calcified plaques. In some cases, such information is useful in determining the treatment plan. In other cases, such information has limited clinical relevance. In some embodiments, non-calcified plaques may include low-density non-calcified plaques, also known as low-attenuation plaques.

[0269] Figure 17 is a table showing an example of a plaque calcification threshold in relation to luminal contrast and kVp. As shown in Figure 17, the plaque calcification threshold generally increases as kVp decreases and generally increases as luminal contrast increases. However, it should be understood that the specific thresholds shown in Figure 17 are for illustrative purposes only, and actual thresholds may vary. In some embodiments, a table like the one shown in Figure 17 can be used as a lookup table to determine the plaque calcification threshold for a given kVp and luminal contrast. In some embodiments, the information shown in such a table can be used to create a multivariable equation describing the relationship between kVp, luminal contrast, and the plaque calcification threshold. In some embodiments, interpolation can be used to determine the threshold for intermediate values ​​of luminal contrast and / or kVp.

[0270] Figure 18 is a flowchart illustrating an exemplary process for determining a threshold HU value to distinguish LAP from other non-calcified plaques, according to several embodiments. The process shown in Figure 18 can be performed by a computer system.

[0271] In operation 1905, the system can receive a set of CCTA images and a set of OCT images. In operation 1910, the system can co-register the CCTA images and OCT images. For example, for each CCTA image, there may be a corresponding OCT image, and the CCTA images and OCT images can be co-registered. In operation 1915, the system can determine the presence of LAP in the OCT image. In operation 1920, the system can determine the presence of LAP in the OCT image. In operation 1925, the system can determine the presence of LAP in the CCTA image. In operation 1930, the system can adjust the CCTA plaque classification threshold so that the determined presence of LAP in the CCTA image is within the threshold of the determined presence of LAP in the corresponding OCT image.

[0272] Figure 19 is a flowchart illustrating an exemplary process for determining a threshold HU value to identify TCFA, according to several embodiments. The process shown in Figure 8 can be performed by a computer system.

[0273] In operation 2005, the system can receive a set of OCT images and a set of CCTA images. In operation 2010, the system can label the OCT images as either TCFA or not TCFA. In some embodiments, the system can analyze each OCT image to determine whether or not it shows signs of TCFA. In some embodiments, the OCT images can be pre-labeled as either showing TCFA or not. Each OCT image can correspond to a CCTA image. For each CCTA image, in operation 2015, the system can determine whether or not the image shows TCFA by, for example, determining whether the LAP in the image extends to the boundary between the plaque core in the image and the lumen in the image. The system can compare this result with the label of the corresponding OCT image and, in operation 2020, determine a threshold. For example, the system can iteratively adjust the threshold. In some embodiments, the system can adjust the threshold until the analysis yields an inaccurate result. In some embodiments, the iterative adjustments can be, for example, 10HU or about 10HU, 20HU or about 20HU, 50HU or about 50HU, or more or less, or any value between these values. In this way, assuming that the distribution of TCFA images and non-TCFA images is approximately equal, the system can determine a range of thresholds for identifying TCFA. In some embodiments, it may not be required that the proportion of images with TCFA and images without TCFA be equal or approximately equal. For example, the results can be weighted so that an overall threshold for kVp and lumen contrast can be determined. In operation 2025, the system can determine thresholds for each combination of kVp and lumen contrast. In some embodiments, lumen contrast can be binned with, for example, 10HU, 20HU, 30HU, 40HU, 50HU, 100HU, or any value between these values, or one or more or less values. Binning can ensure that each combination of kVp and lumen contrast has a sufficient number of samples to determine a reliable threshold.

[0274] In some embodiments, a machine learning model can be trained to identify TCFAs in CCTA images. The machine learning model can be trained using supervised learning as described herein. Figure 20 is a flowchart illustrating an exemplary process for identifying TCFAs in CCTA images using machine learning, according to some embodiments. The process in Figure 20 can be performed by a computer system. In Figure 20, training and deployment are illustrated as a single process. However, it will be understood that training and deployment can be performed as separate processes, either on the same computer system or on different computer systems.

[0275] In operation 2105, the system can receive multiple labeled CCTA images. The labels may indicate whether the CCTA image exhibits TCFA and the kVp used to capture the CCTA image. In some embodiments, the labels may include luminal contrast, while such information may not be included in some other embodiments. In operation 2110, the system can generate a vector representation of the CCTA image. In some embodiments, the vector representation may include one or more labels or representations of parts of one or more labels. For example, the vector representation of the image may encode the image data as well as the kVp used to capture the image. In operation 2115, the system can train a machine learning model using supervised learning. For example, the label "TCFA present" may indicate whether the image exhibits TCFA, and a model can be trained using supervised learning where "TCFA present" is the desired output.

[0276] After training, the model can be deployed. During deployment, the model can receive new CCTA images in which the presence of TCFA is unknown. The presence of TCFA can be determined by the model. In operation 2120, the system can receive a new CCTA image. In operation 2125, the system can generate a vector representation of the received CCTA image. In some embodiments, the vector representation may include kVp values ​​used to capture the CCTA image. In operation 2130, the system can use the machine learning model trained in operation 2115 to determine whether the received CCTA image exhibits TCFA.

[0277] The descriptions of Figures 18, 19, and 20 assume that TCFA identification is independent of the equipment used for imaging. However, this is not always the case. For example, in some embodiments, the X-ray source, X-ray detector, shape (e.g., distance from source to detector), and collimator used can affect the resulting image. In some embodiments, the total radiation dose to which the patient is exposed can affect the resulting image. Often, considering lumen contrast can account for the effects of such variability in image acquisition, but in some cases, including some or all of the above information can improve accuracy.

[0278] Figures 21A and 21B schematically show examples of CCTA images that do not show TCFA (Figure 21A) and CCTA images that show TCFA (Figure 21B). In CCTA image 2300A, the lumen 2302A is surrounded by plaque 2304A. Low-attenuation plaque 2306A is present in some areas. However, the low-attenuation plaque 2306A does not extend to the interface between the lumen and the plaque. In some embodiments, a system implementing one or more of the approaches described herein can identify CCTA image 2300A as not showing TCFA. In CCTA image 2300B, plaque 2304B includes low-attenuation plaque 2306B. In Figure 21B, the low-attenuation plaque 2306B extends to the boundary between the plaque and the lumen 2302B. In some embodiments, a system implementing one or more of the approaches described herein can identify CCTA image 2300B as showing the presence of TCFA.

[0279] Figures 22A and 22B schematically illustrate examples of CCTA images with different LAP cutoff thresholds. In CCTA image 2400A, a low contrast cutoff is used to distinguish LAP 2406A from non-LAP 2404A (which may include non-calcified and / or calcified plaques). CCTA image 2400B schematically shows CCTA image 2400A, but with a different threshold for separating LAP 2406B from other plaques 2404B. As shown in Figures 22A and 22B, the determination of whether TCFA is present (e.g., whether a low-attention plaque reaches the boundary between the plaque and the lumen) may vary depending on the selected threshold.

[0280] Figure 23 shows examples of the separation of thin-capsulated and thick-capsulated fibrous hemangiomas at different Hounsfield unit thresholds in several embodiments. In Figure 23, the apparent capsule thickness using CT may not be able to distinguish between thin-capsulated and thick-capsulated capsules, depending on the selected threshold. For example, at a threshold of 30 HU, thin-capsulated capsules appear thick, and at 75 HU, thick-capsulated capsules appear thin. In the illustrated example, at 60 HU, there is a clear separation between thin-capsulated and thick-capsulated fibrous hemangiomas, and these two can be easily distinguished from each other. It should be understood that Figure 23 is for illustrative purposes only. In practice, actual thresholds may differ from those depicted in Figure 23.

[0281] Figure 24 is a diagram illustrating examples of user interfaces according to several embodiments. A coronary artery tree is illustrated in Figure 24. In some embodiments, anatomical labels may be shown on the coronary artery tree. In some embodiments, parts of the coronary artery tree may be shaded, highlighted, colored, or otherwise marked to display clinically relevant information. For example, in some embodiments, parts of the coronary artery tree may be colored or otherwise distinguished to indicate the level of diametrical stenosis. In some embodiments, areas of chronic total occlusion may be displayed (e.g., by coloring). In some embodiments, areas with stent placement may be indicated. In some embodiments, areas with a high probability of ischemia can be identified. In some embodiments, the user interface may display total atherosclerotic plaque volume, total calcified plaque volume, total non-calcified plaque volume, percentage atheroma volume, and / or any other clinically relevant information. In some embodiments, the user interface may identify vessels with a high probability of ischemia. For example, in Figure 24, the left anterior descending vessel is identified as having a high probability of ischemia. In some embodiments, the user interface can identify severe and / or moderate stenosis. In some embodiments, the user interface can display coronary artery disease-reporting and data system (CAD-RADS) information. In some embodiments, the user interface can include links or other user interface elements for viewing additional details. In some embodiments, the user interface can include selectors for selecting areas (e.g., All, RCA, LM+LAD, Cx, etc.) for visualization and / or analysis.

[0282] Figure 25 is a drawing showing another exemplary user interface according to several embodiments. The user interface may include a linearized view of an artery (LM+LAD in Figure 24). The linearized view may be augmented with various labels (also referred to herein as markers). For example, numerical labels or positional markers may be added to the linearized view to indicate the location of a lesion, e.g., the distance of the lesion from the osstium. In some embodiments, anatomical labels may be added to the linearized view. For example, in Figure 8, labels are provided indicating the left main artery, left anterior descending proximal, left anterior descending medial, and left anterior descending distal. In some embodiments, the labels may indicate the location and value of the maximum diameter stenosis. For example, in Figure 8, for the first lesion, the maximum diameter stenosis of the left main artery is 17%, and the maximum diameter stenosis of the left anterior descending proximal is 53%.

[0283] In some embodiments, the user interface may include a detail view. Segments, average lumen diameter, diameter stenosis, etc., may be provided in the detail view. In some embodiments, the detail view may be shaded, color-coded, or otherwise indicate the level of diameter stenosis.

[0284] Information boxes can be provided in the user interface. These information boxes can provide information about each observed lesion. In some embodiments, the user can click on a lesion in a linearized view or select it in other ways to expand the corresponding information box. In some embodiments, the user can select an information box, and the corresponding lesion can be highlighted or selected in other ways in the linearized view.

[0285] Figure 26 is a flowchart illustrating an exemplary process for generating a surgical plan according to several embodiments. The process shown in Figure 26 can be executed on a computer system. In operation 2802, the system can receive data such as CCTA images, OCT data, IVUS data, and near-infrared spectroscopy data. In operation 2804, the system can generate a graphical user interface showing one or more lesions. In operation 2806, the system can receive a user selection of one or more lesions to be treated. In operation 2808, the system can determine one or more stents. For example, the system can determine the length, diameter, material, and structure of one or more stents. In operation 2810, the system can determine the type of surgery. If the surgery is performed by a human, the system can generate a surgical plan in operation 2812. If the surgery is performed by a robot, the system can generate a robotic surgery plan in 2814.

[0286] [Examples] In one or more embodiments, the technology described herein relates to a computer implementation method for plaque regression classification, the computer implementation method comprising: receiving a first medical image, wherein the first medical image shows a first region of a plaque; receiving a second medical image, wherein the second medical image shows a second region of a plaque; identifying the first region of the plaque in the first medical image; identifying the second region of the plaque in the second medical image; determining a first set of one or more characteristics for the first region of the plaque; determining a second set of one or more characteristics for the second region of the plaque; comparing at least one characteristic of the first set of characteristics with at least one corresponding characteristic of the second set of characteristics; determining a plaque regression between the first medical image and the second medical image based on the comparison; and determining a plaque regression classification based on the comparison.

[0287] In one or more aspects, the technology described herein relates to a computer-implemented method, where a first region of a plaque and a second region of the plaque are the same region of the plaque.

[0288] In one or more aspects, the technology described herein relates to a computer-implemented method, further comprising co-registering a first image and a second image.

[0289] In one or more aspects, the technology described herein relates to a computer-implemented method, where the classification is either regression or pseudo-regression.

[0290] In one or more aspects, the technology described herein relates to a computer-implemented method, where the classification is either true regression, pseudo-regression, stationary, or growing.

[0291] In one or more aspects, the technology described herein relates to a computer-implemented method, where a first set of characteristics includes a first plaque mass, a second set of characteristics includes a second plaque mass, where comparing includes determining a difference between the first plaque mass and the second plaque mass, where pseudo-regression is characterized by the difference being less than a threshold, and where true regression is characterized by the first plaque mass being at least a threshold greater than the second plaque mass.

[0292] In one or more aspects, the technology described herein relates to a computer-implemented method, where the first plaque mass is defined as the product of a first density and a first volume, and the second plaque mass is defined as the product of a second density and a second volume.

[0293] In one or more aspects, the technology described herein relates to a computer-implemented method, where the first density and the second density include a material density or a radiation density.

[0294] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein a first set of one or more characteristics and a second set of one or more characteristics each include at least one of total plaque volume, total plaque area, total plaque length, high-density plaque volume, high-density plaque area, high-density plaque length, low-density plaque volume, low-density plaque area, low-density plaque length, material density, or radiant density.

[0295] In one or more embodiments, the technology described herein relates to a computer mounting method, wherein a first set of one or more properties and a second set of one or more properties each comprise the total plaque mass, where the total plaque mass is determined by the product of the material density and the total plaque amount.

[0296] In one or more embodiments, the technology described herein relates to a computer implementation method, further including displaying classifications to the user via a graphical user interface.

[0297] In one or more embodiments, the technology described herein relates to a computer implementation method for plaque regression classification, the computer implementation method comprising: receiving a first medical image, wherein the first medical image represents a first region of a plaque; receiving a second medical image, wherein the second medical image represents a second region of a plaque; providing representations of the first and second medical images to a machine learning model, wherein the model is configured to analyze the first and second medical images to determine a regression classification; and using the machine learning model to determine a regression classification of the plaque between the first and second medical images.

[0298] In one or more embodiments, the techniques described herein relate to computer implementation methods, wherein a machine learning model is trained using supervised learning on image pairs that exhibit true and pseudo-regression.

[0299] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the first image of a pair of images and the second image of the pair of images represent the same region of the plaque.

[0300] In one or more embodiments, the technology described herein relates to a computer implementation method in which a first image and a second image are captured at different times.

[0301] In one or more embodiments, the technology described herein relates to a computer implementation method in which a first image is captured earlier than a second image.

[0302] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the regression classification is selected from true regression or pseudo-regression.

[0303] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the regression classification is selected from true regression, pseudo-regression, static, or growth.

[0304] In one or more embodiments, the technology described herein relates to a computer implementation method, further comprising displaying a Plaque regression classification to a user via a graphical user interface.

[0305] In one or more embodiments, the technology described herein relates to a computer implementation method for facilitating decisions regarding the treatment of arterial disease in a patient having arterial disease, based at least in part on medical image analysis, wherein the computer implementation method includes: accessing a medical image of the patient by a computer system, wherein the medical image includes a representation of one or more arteries, and the one or more arteries include one or more plaque regions; analyzing the medical image by the computer system to identify one or more arteries and one or more regions of plaque; determining the total amount of plaque by the computer system based at least in part on the plaque in the identified one or more regions; and determining the total amount of plaque in the patient by the computer system. The computer system includes: evaluating the expected therapeutic effect from systemic therapy based at least partially on the quantity; determining the presence and extent of high-risk plaques based at least partially on the density of one or more areas of identified plaques using a computer system; evaluating the expected therapeutic effect from local interventions based at least partially on the presence and extent of high-risk plaques using a computer system; and determining the recommended treatment for arterial disease for the patient based on the expected therapeutic benefits from the evaluated local interventions and the evaluated systemic therapy, wherein the computer system includes a computer processor and an electronic storage medium.

[0306] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the recommended treatment for arterial disease comprises one or more systemic therapies or local interventions, the recommended treatment for arterial disease comprises systemic therapy when the total amount of plaque is greater than or equal to a first predetermined threshold, and the recommended treatment for arterial disease comprises local interventions when the presence and extent of high-risk plaques are greater than or equal to a second predetermined threshold.

[0307] In one or more embodiments, the technology described herein relates to a computer implementation method, and the recommended treatment of arterial disease includes both local treatment and computer system treatment.

[0308] In one or more embodiments, the technology described herein relates to a computer implementation method, further comprising determining an appropriate type or level of systemic treatment, wherein the arterial disease treatment recommended for a patient includes an appropriate type or level of systemic treatment.

[0309] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein one or more of a first predetermined threshold or a second predetermined threshold is determined using a machine learning algorithm.

[0310] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein systemic treatment includes one or more of drug therapy, lifestyle, or dietary therapy.

[0311] In one or more embodiments, the technology described herein relates to a computer implementation method, and the local intervention includes one or more percutaneous coronary interventions (PCI) or coronary artery bypass grafting (CABG).

[0312] In one or more embodiments, the techniques described herein relate to a computer implementation method and further include determining one or more recommended parameters for local treatment.

[0313] In one or more embodiments, the technology described herein relates to a computer implementation method, and one or more recommended parameters include one or more of the dimensions, type, or material of the stent.

[0314] In one or more embodiments, the technology described herein relates to a computer implementation method, and arterial diseases include coronary artery disease (CAD).

[0315] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the medical images include computed tomography (CCTA) angiography images.

[0316] In one or more embodiments, the technology described herein relates to a computer mounting method, wherein one or more densities include material densities.

[0317] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein one or more densities include radiant density.

[0318] In one or more embodiments, the technology described herein relates to a computer implementation method, in which high-risk plaques are determined at the lesion level.

[0319] In one or more embodiments, the technology described herein relates to a computer implementation method, further comprising: generating a patient-level risk score for a patient based at least in part on the determined total plaque volume of the patient; and generating a lesion-level risk score for a patient based at least in part on the presence and extent of determined high-risk plaques, wherein the recommended treatment for arterial disease is determined at least in part on the patient-level risk score and the lesion-level risk score.

[0320] A computer implementation method for normalizing computed tomography (CT) images without using a physical calibration device for the analysis of one or more plaques or vascular parameters, comprising: accessing a patient's CT image by a computer system, wherein the CT image includes representations of one or more arteries, and the one or more arteries include one or more regions of plaque; accessing one or more image acquisition parameters used to acquire the CT image by the computer system, wherein the one or more image acquisition parameters include one or more of the following: helical CT scheme, CT detector type, CT type based on the number of photon energy spectra, current (mA), peak kilovolts (kVp), image noise, signal, signal-to-noise ratio, contrast effect, or contrast-to-noise ratio; and normalizing the CT image by applying an image processing algorithm to the CT image without using a physical calibration device, wherein the image processing algorithm is the same subject The normalized CT images are derived by analyzing multiple test CT images obtained from a person and one or more image acquisition parameters used to obtain the multiple test CT images, wherein the multiple test CT images include one or more arteries containing one or more regions of plaque; wherein the normalized CT images are configured to be analyzed to generate one or more plaque parameters and one or more vascular parameters, wherein the one or more plaque parameters include one or more of total plaque volume, non-calcified plaque volume, or calcified plaque volume, and the one or more vascular parameters include one or more lumen measurements, wherein the one or more plaque parameters and one or more vascular parameters generated from the normalized CT images are configured to be compared with one or more plaque parameters and one or more vascular parameters generated from a second CT image of the patient, wherein the computer system comprises a computer processor and an electronic storage medium.

[0321] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the helical CT scheme includes one or more single-source, dual-source, multi-source, fast-switching, and fast-pitch helical.

[0322] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the type of CT detector includes one or more energy integrating detectors or photon counting detectors.

[0323] In more than one embodiment, the technology described herein relates to a computer implementation method, wherein the type of CT based on the number of photon energy spectra includes one or more of single-energy CT, dual-energy CT, spectral CT, or multispectral CT.

[0324] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein two or more test CT images are acquired using one or more different image acquisition parameters.

[0325] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein two or more test CT images are obtained at two or more different time points.

[0326] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the image processing algorithm includes a polynomial regression equation.

[0327] In one or more embodiments, the technology described herein relates to a computer implementation method, and the image processing algorithm includes a linear regression equation.

[0328] In one or more embodiments, the technology described herein relates to a computer implementation method, and the image processing algorithm includes a machine learning algorithm.

[0329] In one or more embodiments, the technology described herein relates to a computer implementation method, and the image processing algorithm includes a mathematical algorithm.

[0330] In one or more embodiments, the technology described herein relates to a computer implementation method in which one or more image acquisition parameters linearly affect the CT image.

[0331] In one or more embodiments, the technology described herein relates to a computer implementation method in which one or more image acquisition parameters nonlinearly affect the CT image.

[0332] In one or more embodiments, the technology described herein relates to a computer implementation method wherein a second CT image of the patient is obtained at a different time than the first CT image of the patient.

[0333] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein a second CT image of the patient is acquired using one or more different image acquisition parameters from the patient's first CT image.

[0334] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein CT images include coronary computed tomography (CCTA) angiography images.

[0335] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein normalizing CT images includes normalizing a plurality of CT images, and the computer implementation method further includes generating a coronary artery tree based on the normalized CT images, and generating a coronary artery tree includes: extracting coronary arteries from the plurality of normalized CT images by a computer system; labeling the extracted vessels by a computer system; segmenting the patient's aorta present in each of the plurality of normalized CT images by a computer system; ranking each of the plurality of normalized CT images for each extracted vessel by a computer system; selecting an image for each extracted vessel based on the ranking by a computer system; and registering each selected image for generating a tree by a computer system.

[0336] In one or more embodiments, the technology described herein relates to a computer implementation method for generating an image processing algorithm for normalizing computed tomography (CT) images without using a physical calibration device for the analysis of one or more plaques or vascular parameters, wherein the computer implementation method includes: accessing a plurality of test CT images obtained from the same subject by a computer system, wherein the plurality of test CT images include representations of one or more arteries, including one or more regions of atherosclerotic plaques; accessing one or more image acquisition parameters used to obtain the plurality of test CT images by a computer system, wherein the one or more image acquisition parameters include one or more of the helical CT scheme, CT detector type, CT type based on the number of photon energy spectra, current (mA), peak kilovolts (kVp), image noise, signal, signal-to-noise ratio, contrast effect, or contrast-to-noise ratio; and the computer system provides the plurality of test CT images and one or more image acquisition parameters used to obtain the plurality of test CT images. The present invention comprises generating an image processing algorithm for normalizing a patient's CT image without using a physical calibration device, based at least in part on the following: the image processing algorithm is configured to be used to normalize a patient's CT image based at least in part on one or more image acquisition parameters used to acquire the CT image without using a physical calibration device; the normalized CT image is configured to be analyzed to generate one or more plaque parameters and one or more vascular parameters, where the one or more plaque parameters include one or more of total plaque volume, non-calcified plaque volume, or calcified plaque volume; where the one or more vascular parameters include one or more lumen measurements; where the one or more plaque parameters and one or more vascular parameters generated from the normalized CT image are configured to be compared with one or more plaque parameters and one or more vascular parameters generated from another CT image of the patient; and where the computer system includes a computer processor and an electronic storage medium.

[0337] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the helical CT scheme includes one or more single-source, dual-source, multi-source, high-speed switching, or high-speed pitch helical.

[0338] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the type of CT detector includes one or more energy integrating detectors or photon counting detectors.

[0339] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the type of CT based on the number of photon energy spectra includes one or more of single-energy CT, dual-energy CT, spectral CT, or multispectral CT.

[0340] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein two or more test CT images are acquired using one or more different image acquisition parameters.

[0341] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein two or more test CT images are obtained at two or more different time points.

[0342] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the image processing algorithm includes a polynomial regression equation.

[0343] In one or more embodiments, the technology described herein relates to a computer implementation method, and the image processing algorithm includes a regression equation.

[0344] In one or more embodiments, the technology described herein relates to a computer implementation method, and the image processing algorithm includes a machine learning algorithm.

[0345] In one or more embodiments, the technology described herein relates to a computer implementation method, and the image processing algorithm includes a mathematical algorithm.

[0346] In one or more embodiments, the technology described herein relates to a computer implementation method in which one or more image acquisition parameters linearly affect the CT image.

[0347] In one or more embodiments, the technology described herein relates to a computer implementation method in which one or more image acquisition parameters non-linearly affect the CT image.

[0348] In one or more embodiments, the technology described herein relates to a computer implementation method wherein another CT image of the patient is obtained at a different time than the CT image of the patient.

[0349] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein another CT image of a patient is acquired using one or more different image acquisition parameters from the patient's CT image.

[0350] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein CT images include coronary computed tomography (CCTA) angiography images.

[0351] In one or more embodiments, the technology described herein relates to a computer implementation method for risk stratification of coronary artery disease (CAD) based on image analysis of one or more coronary arteries of a subject, wherein the computer implementation method includes: accessing medical images of a subject showing one or more coronary arteries by a computer system; analyzing the medical images by the computer system to determine one or more stenotic sites in one or more coronary arteries; analyzing the medical images by the computer system to determine one or more sites of atherosclerosis in one or more coronary arteries; analyzing the medical images by the computer system to determine one or more ischemic areas in one or more coronary arteries; and generating risk stratification of the subject for major adverse cardiovascular events (MACE) at least partially based on one or more stenotic sites, one or more atherosclerotic sites, and one or more ischemic sites, wherein the computer system includes a computer processor and an electronic storage medium.

[0352] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the risk stratification of a subject's MACE is generated by inputting one or more stenotic areas, one or more atherosclerotic areas, and one or more ischemic areas into a machine learning algorithm.

[0353] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the risk stratification of the subject's MACE includes CAD risk staging.

[0354] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the risk stratification of subjects' MACE is generated at least in part on the risk stratification of subjects' MACE based on one or more stenotic areas, one or more atherosclerotic areas, or one or more ischemic areas identified from one or more medical images collected from multiple subjects.

[0355] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein one or more areas of ischemia in one or more coronary arteries are determined on the basis of at least partially one or more areas of stenosis and one or more areas of atherosclerosis.

[0356] In one or more embodiments, the technology described herein relates to a computer implementation method in which one or more areas of ischemia in one or more coronary arteries are determined using a machine learning algorithm.

[0357] In one or more embodiments, the technology described herein relates to a computer implementation method for constructing a patient's coronary artery tree, the computer implementation method comprising: extracting coronary arteries from a plurality of image series by a computer system; labeling the extracted vessels by a computer system; segmenting the patient's aorta present in each of the plurality of image series by a computer system; ranking each of the plurality of image series for each extracted vessel by a computer system; selecting an image series for each extracted vessel based on the ranking by a computer system; and registering each selected image series for generating a tree by a computer system.

[0358] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the series of images includes a series of computed tomography (CCTA) images.

[0359] In one or more embodiments, the technology described herein relates to a computer implementation method and further includes determining an overlap ratio.

[0360] In one or more embodiments, the technology described herein relates to a computer implementation method, where each of a series corresponds to a cardiac phase.

[0361] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein ranking each series for each extracted vessel includes straightening each extracted vessel.

[0362] In one or more embodiments, the techniques described herein relate to a computer implementation method, which includes ranking each extracted series, providing a machine learning model with linearized extracted blood vessels, and the machine learning model is trained to select the best series from a plurality of series.

[0363] In one or more embodiments, the technology described herein relates to a computer implementation method, and the selection of an optimal series includes selecting a series that minimizes one or more blurs or noises.

[0364] In one or more embodiments, the technology described herein relates to a computer implementation method, and the selection of the best series includes selecting a series that maximizes contrast.

[0365] In one or more embodiments, the technology described herein relates to a computer implementation method, which includes determining a common reference in order to register each image series selected to generate a coronary artery tree.

[0366] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the common reference includes at least a portion of the patient's aorta.

[0367] In one or more embodiments, the technology described herein relates to a computer implementation method, and a common reference includes one or more coronary osteostreaks.

[0368] In one or more embodiments, the techniques described herein relate to a computer implementation method, wherein generating a coronary artery tree further includes suturing multiple segments at multiple branching landmarks.

[0369] In one or more embodiments, the technology described herein relates to a computer implementation method, further comprising overlaying a graphical representation of one or more plaques onto a generated coronary artery tree.

[0370] In one or more embodiments, the techniques described herein relate to a computer implementation method, further comprising performing fractional flow reserve-computed tomography (FFR-CT) calculations based at least partially on a generated coronary artery tree.

[0371] In one or more embodiments, the techniques described herein relate to a computer implementation method and further include generating a visualization of a cardiomyocyte at risk based at least partially on a generated coronary artery tree.

[0372] In one or more embodiments, the techniques described herein relate to computer implementation methods and further include simulating a virtual myocardial perfusion map.

[0373] In one or more embodiments, the techniques described herein relate to computer implementation methods and further include simulating quantitative myocardial blood flow of the heart.

[0374] In one or more embodiments, the technology described herein relates to a computer implementation method, the computer implementation method comprising: receiving a first image set; receiving a second image set, wherein the second image set comprises a plurality of coronary computed tomography angiography images, the second image set has a plurality of associated peak kilovolts, the second image set has a plurality of associated lumen contrast values, and each image in the second image set is associated with a corresponding image in the first image set to form a plurality of image pairs; determining a first size of calcified plaque in each first image of the first image set; and determining a second size of calcified plaque in each second image of the second image set. The method includes: determining a second size of calcified plaque in an image; for each pair of multiple image pairs: adjusting the plaque calcification threshold of the second image of the image pair, where adjusting the plaque calcification threshold changes the determined second size of the calcified plaque in the second image; determining the lumen contrast related to the lumen of the vessel in the second image; and determining the peak kilovolts related to the second image; and determining the resulting plaque calcification threshold for each combination of lumen contrast and peak kilovolts related to the second set of images, based on the lumen contrast, peak kilovolts, and plaque calcification threshold of each second image.

[0375] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the first size and the second size include at least one of the amount, area, or thickness of the calcified plaque.

[0376] In one or more embodiments, the techniques described herein relate to a computer implementation method, wherein determining the resulting plaque calcification threshold for each combination of lumen contrast and peak kilovolts includes determining the mean plaque calcification threshold for each combination of lumen contrast and peak kilovolts.

[0377] In one or more embodiments, the techniques described herein relate to a computer implementation method, wherein determining the resulting plaque calcification threshold for each combination of luminal contrast and peak kilovolts includes removing one or more abnormal plaque calcification thresholds.

[0378] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein a first set of images includes a set of near-infrared X-ray spectroscopic images.

[0379] In one or more embodiments, the techniques described herein relate to a computer implementation method, further comprising determining an equation configured to output a calculated plaque calcification threshold for a given luminal contrast and peak kilovolt, based on the results of the plaque calcification threshold for each combination of luminal contrast and peak kilovolt.

[0380] In one or more embodiments, the technology described herein relates to a computer-implemented method for determining the size of calcified plaques, the computer-implemented method comprising: receiving a computed tomography angiography image; identifying a lumen in the computed tomography angiography image; determining the lumen contrast of the lumen; determining the peak kilovolts associated with the computed tomography image; adjusting a plaque calcification threshold for distinguishing calcified and non-calcified plaques, wherein the adjustment is based on a plaque calcification threshold determined by analyzing a plurality of image pairs, wherein each image pair comprises a computed tomography angiography image and a ground truth image; and determining the calcified plaque size, wherein determining the calcified plaque size includes determining the amount of plaque greater than or equal to the plaque calcification threshold.

[0381] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the calcified plaque size is a volume, and the volume is determined by determining the volume of plaques that are greater than or equal to a plaque calcification threshold.

[0382] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the calcified plaque size is an area, and the area is determined by determining the area of ​​plaques that are greater than or equal to a plaque calcification threshold.

[0383] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the calcified plaque size is thickness, and the thickness is determined by the length of the plaque being greater than or equal to the plaque calcification threshold.

[0384] In one or more embodiments, the technology described herein relates to a computer implementation method, the computer implementation method comprising: receiving a first image set; receiving a second image set, wherein the second image set comprises a plurality of coronary computed tomography (CCTA) images, the second image set has a plurality of associated peak kilovolts, the second image set has a plurality of associated lumen contrast values, and each image in the second image set is associated with a corresponding image in the first image set to form a plurality of image pairs; determining the presence of fibrous tissue dysplasia (TCFA) for each first image in the first image set; and the second image set The method includes: determining the proximity of the low-attention plaque (LAP) to the boundary between the plaque and the lumen for each second image of a set of images; adjusting the LAP threshold of the second image of the image pair, where adjusting the LAP threshold changes the proximity of the LAP to the boundary; determining the lumen contrast related to the lumen of the second image; determining the peak kilovolts related to the second image; and determining the resulting LAP threshold for each combination of lumen contrast and peak kilovolts related to the second set of images, based on the lumen contrast, peak kilovolts, and LAP threshold of each second image.

[0385] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the first image set includes a plurality of optical coherence tomography images.

[0386] In one or more embodiments, the techniques described herein relate to a computer implementation method, wherein determining the resulting LAP threshold for each combination of lumen contrast and peak kilovolts includes determining the mean plaque calcification threshold for each combination of lumen contrast and peak kilovolts.

[0387] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein determining the resulting LAP threshold for each combination of luminal contrast and peak kilovolts includes removing one or more outlier LAP thresholds.

[0388] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein determining the resulting LAP threshold for each combination of lumen contrast and peak kilovolts includes binning the peak kilovolts.

[0389] In one or more embodiments, the technology described herein relates to a computer mounting method, wherein the bin size is 10HU, 20HU, 30HU, 40HU, 50HU, 60HU, 70HU, 80HU, 90HU, or 100HU.

[0390] In one or more embodiments, the techniques described herein relate to a computer-implemented method for identifying microfibromas (TCFAs), the computer-implemented method comprising: receiving coronary computed tomography (CCTA) images by a computer system; identifying lumen contrast in the CCTA images by a computer system; identifying peak kilovolts associated with the CCTA images by a computer system; adjusting a low-attenuation plaque (LAP) threshold based on lumen contrast and peak kilovolts by a computer system; identifying low-attenuation plaques adjacent to the lumen by a computer system; and determining the presence of TCFAs based on proximity by a computer system.

[0391] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the LAP threshold is determined by: receiving a first set of images by a computer system; receiving a second set of images by a computer system, wherein the second set of images comprises a plurality of coronary computed tomography angiography images, the second set of images has a plurality of associated peak kilovolts, the second set of images has a plurality of associated lumen contrast values, each image in the second set of images is associated with a corresponding image in the first set of images, forming a plurality of image pairs; and for each first image in the first set of images, the first size of the calcified plaque in the first image is determined To determine the size; for each second image in the second image set, to determine the second size of the calcified plaque in the second image; for each pair of multiple image pairs: to adjust the plaque calcification threshold of the second image of the image pair, where adjusting the plaque calcification threshold changes the determined second size of the calcified plaque in the second image; to determine the lumen contrast related to the vascular lumen in the second image; to determine the peak kilovolts related to the second image; and to determine the LAP threshold for each combination of lumen contrast and peak kilovolts related to the second image set, based on the lumen contrast, peak kilovolts, and plaque calcification threshold of each second image.

[0392] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the second image set has a plurality of associated peak kilovolts.

[0393] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein a second set of images has a plurality of associated lumen contrast values.

[0394] In one or more embodiments, the techniques described herein relate to a computer implementation method for training a machine learning model for identifying microfibromas (TCFAs), the computer implementation method comprising: receiving a set of coronary computed tomography angiography images, wherein each image has a peak kilovolt and a label indicating the presence of a TCFA associated with it; and training a machine learning model for identifying TCFAs using supervised learning, wherein the machine learning model is trained to accept CCTA images and peak kilovolts as input and to output the probability that a CCTA image indicates a TCFA, the training comprising adjusting one or more weights associated with the machine learning model.

[0395] In one or more embodiments, the techniques described herein relate to a computer implementation method for identifying fibrous hemangiomas with thinning hair (TCFA) using a machine learning model, comprising: a computer system receiving computed tomography (CCTA) images; a computer system receiving peak kilovolt values ​​associated with the CCTA images; a computer system generating representations of the CCTA images and peak kilovolts; a computer system applying a machine learning model to the representations; and a computer system determining the output of the machine learning model, wherein the output indicates the probability that the CCTA image indicates a TCFA; wherein the machine learning model is trained by: receiving a set of computed tomography images, where each image has a peak kilovolt and a label indicating the presence of a TCFA associated with it; and performing machine learning to identify TCFAs using supervised learning, wherein the machine learning model is trained to accept CCTA images and peak kilovolts and output the probability that the CCTA image indicates a TCFA, and the training includes adjusting one or more weights associated with the machine learning model.

[0396] In one or more embodiments, the technology described herein relates to a computer implementation method, further comprising a computer system determining that the probability is above a threshold and displaying to the user an indication that TCFA is likely to occur.

[0397] In one or more embodiments, the technology described herein relates to a computer implementation method, further comprising a computer system determining that the probability is below a threshold and displaying to the user an indication that TCFA is unlikely to occur.

[0398] In one or more embodiments, the technology described herein relates to a computer implementation method, further comprising displaying probabilities to a user.

[0399] In one or more embodiments, the technology described herein relates to a computer implementation method for providing a graphical user interface for coronary artery analysis, the computer implementation method comprising: receiving coronary artery data, wherein the coronary artery data comprises linearized computed tomography angiography images and data relating to one or more lesions, where the lesions correspond to one or more segments of a vessel in which plaque exists; generating a graphical user interface, wherein the graphical user interface comprises linearized computed tomography images; one or more information boxes, each information box corresponding to a lesion among the one or more lesions; one or more position markers indicating the beginning and end of each lesion among the one or more lesions; one or more labels, where each label corresponds to one or more lesion portions, and each label is positioned relative to the corresponding lesion portion as shown in the linearized computed tomography images; and providing a graphical user interface for display on a user's computer system.

[0400] In one or more embodiments, the technology described herein relates to a computer implementation method, the computer implementation method further includes: receiving a selection of one or more lesions by a user; and extending a corresponding information box to display data related to the lesion, wherein the user selects a lesion by selecting a lesion shown in a linearized coronary computed tomography image or by selecting one or more information boxes.

[0401] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the data relating to one or more lesions includes at least one of the following: maximum diameter stenosis, minimum lumen diameter stenosis, plaque length, plaque volume, or normal lumen diameter.

[0402] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the graphical user interface further includes one or more markers, each marker positioned relative to a linearized coronary computed tomography angiography image to indicate the location of maximum diameter stenosis.

[0403] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the graphical user interface further includes a scale indicating the distance from a reference point in a linearized coronary computed tomography angiography image.

[0404] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the reference point is the periosteum.

[0405] In one or more embodiments, the technology described herein relates to a computer implementation method, the graphical user interface further includes a display of the possibility of ischemia.

[0406] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the graphical user interface further includes one or more labels corresponding to one or more anatomical features shown in a linearized coronary computed tomography angiography image.

[0407] In one or more embodiments, the technology described herein relates to a computed tomography implementation method, wherein the coronary artery data further comprises at least one of optical coherence tomography data, near-field infrared spectroscopy data, ultrasound data, or invasive angiography data.

[0408] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein data relating to one or more lesions include the length of the plaque and the normal lumen diameter, the distance from a reference point is used to determine the position for stent placement, and the length of the plaque and the normal lumen diameter are used to determine the length and diameter for the stent.

[0409] In one or more embodiments, the technology described herein relates to a computer implementation method, the computer implementation method further includes: The system includes receiving from the user the selection of one or more lesions for intervention, wherein the intervention comprises placing one or more stents in each of the one or more lesions; generating a surgical plan based on the selected one or more lesions, wherein the surgical plan indicates the location where each stent will be placed, the location corresponds to a distance from a reference point, the surgical plan indicates the diameter of each stent, and the surgical plan indicates the length of each stent.

[0410] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the surgical plan is a robotic surgical plan that includes instructions for the movement of a surgical robot.

[0411] In one or more embodiments, the technology described herein relates to a computer implementation method, wherein the reference point is selected from the periosteum, the left main trifractal, or the left main bifractal.

[0412] [Conclusion] In conclusion, the systems and processes described in the aforementioned specification have been explained with reference to specific embodiments thereof. However, it will be apparent that various modifications and changes are possible without departing from the broader spirit and scope of the embodiments disclosed herein. Accordingly, this specification and the drawings are to be considered illustrative rather than restrictive.

[0413] In fact, while systems and processes have been disclosed in the context of specific embodiments and examples, it will be understood by those skilled in the art that various embodiments of systems and processes extend beyond the specifically disclosed embodiments to other alternative embodiments and / or uses of systems and processes, as well as obvious modifications and equivalents thereof. In addition, while some variations of embodiments of systems and processes have been shown and described in detail, other modifications that fall within the scope of this disclosure will be readily apparent to those skilled in the art based on this disclosure. Furthermore, various combinations or subcombinations of specific features and aspects of embodiments are intended to remain within the scope of this disclosure. It should be understood that various features and aspects of disclosed embodiments can be combined or substituted for each other to form various aspects of the disclosed embodiments of systems and processes. The methods disclosed herein do not need to be performed in the order described. Accordingly, it is intended that the scope of systems and processes disclosed herein should not be limited by the specific embodiments described above.

[0414] The systems and methods of this disclosure each have several innovative aspects, and it will be understood that no single one is solely responsible for or required for the desirable attributes disclosed herein. The various features and processes described above may be used independently of each other or combined in various ways. All possible combinations and subcombinations are intended to fall within the scope of this disclosure.

[0415] Certain features described herein in the context of separate embodiments may also be implemented in combination in a single embodiment. Conversely, various features described in the context of a single embodiment may also be implemented separately or in any suitable subcombination in multiple embodiments. Furthermore, features are described above as acting in a particular combination, and may even be initially claimed as such; however, one or more features from the claimed combination may be removed from the combination, and the claimed combination may be directed towards a subcombination or a variation of a subcombination. A single feature or group of features is not required or essential to each embodiment.

[0416] Furthermore, conditional language used herein, particularly "can," "may," "might," "possibly," and "for example," will generally be understood to convey that a particular embodiment includes certain features, elements, and / or operations, while other embodiments do not, unless otherwise stated or understood in the context in which they are used. Therefore, such conditional statements are not generally intended to mean that features, elements, and / or operations are required in some way to one or more embodiments, or that one or more embodiments necessarily include logic for determining, with or without author input or prompting, whether these features, elements, and / or operations are included in or performed in any particular embodiment. Terms such as "comprising," "including," and "having" are synonymous, used comprehensively and in an open-ended manner, and do not exclude additional elements, features, actions, operations, etc. Furthermore, the term "or" is used in a comprehensive (not exclusive) sense; for example, when used to connect a list of elements, "or" means one, some, or all of the elements in the list. Furthermore, the articles “a,” “an,” and “the” used in this application and the attached claims shall be interpreted as meaning “one or more” or “at least one” unless otherwise specified. Similarly, while operations may be depicted in a particular order in the drawings, it should be recognized that such operations do not need to be performed in the specific order shown, nor in a sequential order, nor do all illustrated operations need to be performed in order to achieve the desired result. Furthermore, drawings may schematically illustrate one or more exemplary steps in the form of a flowchart. However, other operations not illustrated may be incorporated into the schematicly illustrated exemplary methods and processes. For example, one or more additional operations may be performed before, after, simultaneously with, or between any of the illustrated operations. Furthermore, in other embodiments, operations may be rearranged or changed in order.In certain situations, multitasking and parallel processing may be advantageous. Furthermore, the separation of various system components in the embodiments described above should not be understood as necessary in all embodiments, and the described program components and systems may generally be integrated together in a single software product or packaged in multiple software products. Further embodiments are within the scope of the following claims. In some cases, the operations described in the claims may achieve the desired results even if they are executed in a different order.

[0417] Furthermore, the methods and apparatus described herein may be influenced by various modifications and alternative forms, specific examples of which are shown in the drawings and described in detail herein. However, it should be understood that embodiments are not limited to any particular form or method disclosed; rather, embodiments cover all modifications, equivalents, and alternatives that fall within the spirit and scope of the various embodiments and appended claims described herein. Furthermore, any disclosure herein of certain features, aspects, methods, characteristics, qualities, attributes, elements, etc., relating to a particular embodiment or form may be used in all other embodiments or forms specified herein. The methods disclosed herein do not need to be performed in the order described herein. The methods disclosed herein may include certain actions performed by the practitioner, but the methods may also include any third-party instructions regarding those actions, explicitly or implicitly. Also, the scope disclosed herein includes all overlaps, partial scopes, and combinations thereof. Expressions such as “up to,” “at least,” “greater than,” “less than,” and “between” include the numbers described. Furthermore, numerical values ​​preceded by terms such as "about" or "approximately" should be interpreted in context, including the stated value (e.g., as accurately as reasonably possible under the circumstances, e.g., ±5%, ±10%, ±15%). For example, "about 3.5 mm" includes "3.5 mm". Phrases preceded by terms such as "substantially" should be interpreted in context, including the mentioned phrase (e.g., as accurately as reasonably possible under the circumstances). For example, "substantially constant" includes "constant". Unless otherwise specified, all measurements are taken under standard conditions, including temperature and pressure.

[0418] Where used herein, the phrase “at least one” in a list of items refers to any combination of those items, including a single member. For example, “at least one of A, B, or C” is intended to include A, B, or C; “at least one of A, B, and C” is intended to cover A, B, C, A, and C; conjunctions such as “at least one of X, Y, and Z” are understood in context as commonly used to indicate that an item, term, etc., may be at least one of X, Y, or Z, unless otherwise noted. Where headings are provided herein, they are for convenience only and do not necessarily affect the scope or meaning of the apparatus and methods disclosed herein.

[0419] Accordingly, the claims are not intended to be limited to the embodiments shown herein, but rather to be the broadest scope consistent with the present disclosure, the principles and novel features disclosed herein.

Claims

1. A computer implementation method for normalizing computed tomography (CT) images without using a physical calibration device for the analysis of one or more plaques or vascular parameters, A computer system accesses the patient's CT images, wherein the CT images include representations of one or more arteries, and each of these arteries includes one or more regions of plaque; The computer system accesses one or more image acquisition parameters used to acquire a CT image, wherein the one or more image acquisition parameters include one or more of the following: helical CT scheme, CT detector type, CT type based on the number of photon energy spectra, current (mA), peak kilovolts (kVp), image noise, signal, signal-to-noise ratio, contrast effect, or contrast-to-noise ratio; A computer system normalizes CT images by applying an image processing algorithm to them without using a physical calibration device, wherein the image processing algorithm is derived by analyzing multiple test CT images obtained from the same subject and one or more image acquisition parameters used to obtain the multiple test CT images, wherein the multiple test CT images include one or more arteries containing one or more regions of plaque; Includes, Here, the normalized CT image is configured to be analyzed to generate one or more plaque parameters and one or more vascular parameters, wherein the one or more plaque parameters include one or more of total plaque volume, non-calcified plaque volume, or calcified plaque volume, and the one or more vascular parameters include one or more lumen measurements. Here, the one or more plaque parameters and the one or more vascular parameters generated from the normalized CT image are configured to be compared with the one or more plaque parameters and the one or more vascular parameters generated from the patient's second CT image. Here, the computer system is a computer implementation method comprising a computer processor and an electronic storage medium.

2. The computer implementation method according to claim 1, wherein the helical CT system comprises one or more of the following: single source, dual source, multi source, fast switching, and fast pitch helical.

3. The computer implementation method according to claim 1, wherein the type of CT detector comprises one or more energy integrating detectors or photon counting detectors.

4. The computer implementation method according to claim 1, wherein the type of CT based on the number of photon energy spectra is one or more of single-energy CT, dual-energy CT, spectral CT, and multispectral CT.

5. The computer implementation method according to claim 1, wherein the plurality of test CT images are acquired using one or more different image acquisition parameters.

6. The computer implementation method according to claim 1, wherein the plurality of test CT images are obtained at two or more different time points.

7. The computer implementation method according to claim 1, wherein the image processing algorithm comprises a multinomial regression equation.

8. The computer implementation method according to claim 1, wherein the image processing algorithm comprises a linear regression equation.

9. The computer implementation method according to claim 1, wherein the image processing algorithm comprises a machine learning algorithm.

10. The computer implementation method according to claim 1, wherein the second CT image of the patient is obtained at a different time than the first CT image of the patient.

11. The computer implementation method according to claim 1, wherein the second CT image of the patient is acquired using one or more different image acquisition parameters from the patient's first CT image.

12. The computer implementation method according to claim 1, wherein the CT image comprises a coronary computed tomography (CCTA) angiography image.

13. Normalizing CT images includes normalizing multiple CT images. The computer implementation method further includes generating a coronary artery tree based on normalized CT images, Generating the aforementioned coronary artery tree is The computer system extracts coronary arteries from multiple normalized CT images; The computer system labels the extracted blood vessels; The computer system segments the patient's aorta present in each of multiple normalized CT images; The computer system ranks each image from multiple normalized CT images for each extracted blood vessel; The computer system selects images for each blood vessel extracted based on rankings; The computer system registers each selected image in order to generate a tree, The computer implementation method according to claim 1, including the method described in claim 1.

14. It is a system, It comprises at least one hardware processor and at least one non-temporary memory for storing instructions, When the instruction is executed by the at least one hardware processor, it causes the system to do the following: Accessing a patient's CT image, wherein the CT image includes a representation of one or more arteries, and the one or more arteries include one or more regions of plaque; Accessing one or more image acquisition parameters used to acquire a CT image, wherein the one or more image acquisition parameters include one or more of the following: helical CT scheme, CT detector type, CT type based on the number of photon energy spectra, current (mA), peak kilovolts (kVp), image noise, signal, signal-to-noise ratio, contrast effect, or contrast-to-noise ratio; Normalizing CT images by applying an image processing algorithm to CT images without using a physical calibration device, wherein the image processing algorithm is derived by analyzing multiple test CT images obtained from the same subject and one or more image acquisition parameters used to obtain the multiple test CT images, wherein the multiple test CT images include one or more arteries containing one or more regions of plaque; Here, the normalized CT image is configured to be analyzed to generate one or more plaque parameters and one or more vascular parameters, wherein the one or more plaque parameters include one or more of total plaque volume, non-calcified plaque volume, or calcified plaque volume, and the one or more vascular parameters include one or more lumen measurements. Herein, the system is configured such that the one or more plaque parameters and the one or more vascular parameters generated from a normalized CT image are compared with the one or more plaque parameters and the one or more vascular parameters generated from a second CT image of the patient.

15. The system according to claim 14, wherein the helical CT method comprises one or more of single-source, dual-source, multi-source, fast-switching, and fast-pitch helical.

16. The system according to claim 14, wherein the type of CT detector comprises one or more energy integrating detectors or photon counting detectors.

17. The system according to claim 14, wherein the type of CT based on the number of photon energy spectra is one or more of single-energy CT, dual-energy CT, spectral CT, and multispectral CT.

18. The system according to claim 14, wherein the plurality of test CT images are acquired using one or more different image acquisition parameters.

19. The system according to claim 14, wherein the plurality of test CT images are obtained at two or more different time points.

20. The system according to claim 14, wherein the CT image comprises a coronary computed tomography (CCTA) angiography image.

Images