Image analysis method, system and storage medium

Abstract

An embodiment of the present specification provides a method for image analysis, which includes acquiring multiple image sequences of a target at a current time point; acquiring multiple image sequences of one or more historical time points of the target; aligning the multiple image sequences including the current time point and each of the one or more historical time points to obtain an image sequence alignment result for each time point; determining a reconstructed image for each time point based on the image sequence alignment result; performing vascular plaque analysis based on the reconstructed image to determine newly added plaques and / or historical plaques, and performing historical plaque analysis on the historical plaques; and determining changes in vascular plaques based on the historical plaque analysis.

Description

Technical Field

[0001] This specification relates to the technical field of medical image analysis, and in particular to an image analysis method, system, and storage medium. Background Art

[0002] Analysis of vascular plaque changes can be performed using ultrasound, but ultrasound examinations are not reproducible, cannot examine intracranial conditions, and cannot directly compare information from multiple time points. Analysis of vascular plaque changes can also be performed using other examination methods (such as CT), but these methods require a certain radiation dose and are not suitable for analyzing high-frequency changes. CT is also insensitive to detecting vessel walls and soft plaques.

[0003] Therefore, it is necessary to provide an image analysis method and system that can ensure the accuracy of the obtained vascular plaque changes without harming the body. Summary of the Invention

[0004] One embodiment of this specification provides an image analysis method. The image analysis method includes: acquiring multiple image sequences of a target at a current time point; acquiring multiple image sequences of the target at one or more historical time points; aligning the multiple image sequences including the current time point and each of the one or more historical time points to obtain an image sequence alignment result for each time point; determining a reconstructed image for each time point based on the image sequence alignment result; performing vascular plaque analysis based on the reconstructed image to determine newly added plaques and / or historical plaques, and performing historical plaque analysis on the historical plaques; and determining vascular plaque changes based on the historical plaque analysis.

[0005] In some embodiments, the changes in the vascular plaque include one or more of changes in plaque size, changes in plaque composition, and changes in the degree of stenosis.

[0006] In some embodiments, the plurality of image sequences includes dark blood images and bright blood images.

[0007] In some embodiments, the plurality of image sequences include a T1 sequence, a T1CE sequence, and a TOF sequence; or, the plurality of image sequences include a T1 sequence, a T1CE sequence, and a CEMRA sequence.

[0008] In some embodiments, the method further includes: predicting the vascular plaque condition at a future time point through a machine learning model based on the changes in the vascular plaque.

[0009] One of the embodiments of the present specification provides an image analysis system, which includes: a first acquisition module for acquiring multiple image sequences of a target at a current time point; a second acquisition module for acquiring multiple image sequences of one or more historical time points of the target; an alignment module for aligning the multiple image sequences including the current time point and each of the one or more historical time points, and obtaining an alignment result of the image sequence at each time point; a reconstruction module for determining a reconstructed image at each time point based on the image sequence alignment result; an analysis module for performing vascular plaque analysis based on the reconstructed image, determining newly added plaques and / or historical plaques, and performing historical plaque analysis on the historical plaques; and a determination module for determining changes in vascular plaques based on the historical plaque analysis.

[0010] One of the embodiments of this specification provides an image analysis device, which includes a processor and a memory; the memory is used to store instructions, and when the instructions are executed by the processor, the device implements the image analysis method.

[0011] One embodiment of this specification provides a computer-readable storage medium, wherein the storage medium stores computer instructions. When a computer reads the computer instructions in the storage medium, the computer executes the image analysis method.

[0012] To address the problem of accurately assessing changes in vascular plaque without harming the body, this invention uses a magnetic resonance imaging system to align, reconstruct, and analyze image sequences at current and historical time points. This allows for quantitative determination of plaque changes, resulting in a more accurate picture of plaque changes. Furthermore, using a machine learning model to predict plaque conditions at future time points provides a better understanding of the future development of plaques. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] This specification will be further described in the form of exemplary embodiments, which will be described in detail with reference to the accompanying drawings. These embodiments are not limiting, and in these embodiments, like numbers represent like structures, wherein:

[0014] Figure 1 is a schematic diagram of an application scenario of an image analysis system according to some embodiments of this specification;

[0015] Figure 2 is a system module diagram of image analysis according to some embodiments of this specification;

[0016] Figure 3 is an exemplary flow chart of image analysis according to some embodiments of this specification;

[0017] Figure 4 is a schematic diagram of a machine learning model according to some embodiments of this specification;

[0018] Figure 5 is a schematic diagram of machine learning model training according to some embodiments of this specification;

[0019] Figure 6A is a black blood image according to some embodiments of this specification;

[0020] Figure 6B is a bright blood image according to some embodiments of this specification;

[0021] Figure 7A is a schematic diagram of a CPR reconstructed image according to some embodiments of this specification;

[0022] Figure 7B is a schematic diagram of centerline extraction according to some embodiments of this specification;

[0023] Figure 7C is a schematic diagram of an MPR reconstructed image according to some embodiments of this specification. DETAILED DESCRIPTION

[0024] To more clearly illustrate the technical solutions of the embodiments of this specification, the following briefly describes the drawings required for describing the embodiments. Obviously, the drawings described below are merely examples or embodiments of this specification. Those skilled in the art can apply this specification to other similar scenarios based on these drawings without inventive effort. Unless otherwise apparent from the context or otherwise noted, the same reference numerals in the figures represent the same structure or operation.

[0025] It should be understood that the terms "system," "device," "unit," and / or "module" used herein are a method for distinguishing different components, elements, parts, portions, or assemblies at different levels. However, if other terms can achieve the same purpose, the terms may be replaced by other expressions.

[0026] As used in this specification and claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" do not refer to the singular but also include the plural. Generally speaking, the terms "comprises" and "include" only indicate the inclusion of the steps and elements specifically identified, and these steps and elements do not constitute an exclusive list. A method or apparatus may also include other steps or elements.

[0027] Flowcharts are used throughout this specification to illustrate the operations performed by systems according to embodiments of this specification. It should be understood that preceding or following operations do not necessarily need to be performed in exact order. Instead, the steps may be processed in reverse order or simultaneously. Furthermore, other operations may be added to these processes, or one or more operations may be removed from these processes.

[0028] Figure 1 1 is a schematic diagram of an application scenario of an image analysis system according to some embodiments of this specification. In some embodiments, image analysis scenario 100 may include a magnetic resonance (MR) system. In some embodiments, image analysis scenario 100 may include modules and / or components for performing image analysis.

[0029] Just as an example, Figure 1 As shown, the image analysis scenario 100 may include an image analysis device 110 , a processing device 120 , a storage device 130 , a terminal 140 , and a network 150 .

[0030] The image analysis device 110 may include an imaging device, an interventional medical device, or a combination thereof. The imaging device may acquire a reconstructed image related to at least a portion of an object. The object may be biological. For example, the object may include a carotid artery, an upper limb artery, a lower limb artery, a thoracic aorta, an abdominal aorta, a coronary artery, a cerebral blood vessel, or any combination thereof. Exemplary imaging devices may include an MRI scanner, a PET scanner, a CT scanner, a DR scanner, or any combination thereof. Exemplary interventional medical devices may include radiotherapy (RT) equipment, ultrasound therapy equipment, thermal therapy equipment, surgical intervention equipment, or any combination thereof.

[0031] The processing device 120 may process data and / or information obtained from the image analysis device 110, the terminal 140, and / or the storage device 130. For example, the processing device 120 may perform image analysis by processing a reconstructed image acquired by the image analysis device 110.

[0032] In some embodiments, processing device 120 may perform image analysis based on one or more models corresponding to image analysis. For example, processing device 120 may determine a reconstructed image at each time point based on the image sequence alignment results at each time point. As another example, processing device 120 may perform vascular plaque analysis based on the reconstructed image. As yet another example, processing device 120 may determine vascular plaque changes based on the vascular plaque analysis.

[0033] The storage device 130 may store data, instructions, and / or any other information. In some embodiments, the storage device 130 may store data obtained from the terminal 140 and / or the processing device 120. The data may include reconstructed images acquired by the processing device 120, models used to process the reconstructed images, information related to components of the processing device 120, and the like. For example, the storage device 130 may store a reconstructed image determined by the alignment results of an image sequence at each time point. As another example, the storage device 130 may store one or more models used to process the reconstructed images.

[0034] In some embodiments, the storage device 130 may be connected to the network 150 to communicate with one or more other components of the image analysis scenario 100 (e.g., the processing device 120, the terminal 140, etc.). One or more components of the image analysis scenario 100 may access data or instructions stored in the storage device 130 through the network 150. In some embodiments, the storage device 130 may be part of the processing device 120.

[0035] The terminal 140 may include a mobile device 140 - 1 , a tablet computer 140 - 2 , a laptop computer 140 - 3 , etc., or any combination thereof. In some embodiments, the terminal 140 may be part of the processing device 120 .

[0036] In some embodiments, terminal 140 can send and / or receive information related to image analysis to processing device 120 via a user interface. In some embodiments, the user interface can be in the form of an application for image analysis implemented on terminal 140. The user interface can be configured to facilitate communication between terminal 140 and a user associated with terminal 140. In some embodiments, the user interface can receive input from a user requesting to perform image analysis, for example, via a user interface screen. Terminal 140 can send a request to perform image analysis to processing device 120 via the user interface in order to analyze a vascular plaque image.

[0037] The network 150 may include any suitable network that can facilitate information and / or data exchange of the image analysis scenario 100. In some embodiments, one or more components of the image analysis device 110 (e.g., an MRI scanner, a CT scanner, a PET scanner, etc.), the terminal 140, the processing device 120, the storage device 130, etc. can communicate information and / or data with one or more other components of the image analysis scenario 100 via the network 150.

[0038] It should be noted that the above description of the image analysis scene 100 is provided for illustrative purposes only and is not intended to limit the scope of the present application. A person of ordinary skill in the art may make various changes and modifications based on the teachings of this application. For example, the assembly and / or functionality of the image analysis scene 100 may be varied or modified according to the specific implementation. By way of example only, other components may be added to the image analysis scene 100, such as a power module that can supply power to one or more components of the image analysis scene 100, as well as other devices or modules.

[0039] Figure 2 It is a system module diagram of image analysis according to some embodiments of this specification.

[0040] In some embodiments, the image analysis system 200 may include a first acquisition module 210 , a second acquisition module 220 , an alignment module 230 , a reconstruction module 240 , an analysis module 250 , and a determination module 260 .

[0041] In some embodiments, the first acquisition module 210 may be configured to acquire a plurality of image sequences of the target at a current time point.

[0042] In some embodiments, the second acquisition module 220 may be configured to acquire a plurality of image sequences of the target at one or more historical time points.

[0043] In some embodiments, the alignment module 230 may be configured to align the multiple image sequences including the current time point and each of the one or more historical time points, and obtain an image sequence alignment result for each time point.

[0044] In some embodiments, the reconstruction module 240 may be configured to determine the reconstructed image at each time point based on the image sequence alignment result.

[0045] In some embodiments, the analysis module 250 may be configured to perform vascular plaque analysis based on the reconstructed image, determine newly added plaques and / or historical plaques, and perform historical plaque analysis on the historical plaques.

[0046] In some embodiments, the determination module 260 may be configured to determine changes in vascular plaque based on the historical plaque analysis.

[0047] In some embodiments, changes in vascular plaques may include one or more of changes in plaque size, plaque composition, and stenosis severity. In some embodiments, the multiple image sequences may include black blood images and bright blood images. In some embodiments, the multiple image sequences may include a T1 sequence, a T1CE sequence, and a TOF sequence; or, the multiple image sequences may include a T1 sequence, a T1CE sequence, and a CEMRA sequence.

[0048] In some embodiments, the image analysis system 200 may further include a prediction module (not shown) that can be used to predict the vascular plaque condition at a future time point based on the vascular plaque change condition using a machine learning model.

[0049] Figure 3 FIG. 1 is an exemplary flow chart of image analysis according to some embodiments of this specification. Figure 3 As shown, process 300 includes the following steps. In some embodiments, Figure 3 One or more operations of the illustrated process 300 may be performed in Figure 1 The image analysis scenario 100 is shown. For example, Figure 3 The illustrated process 300 may be stored in the form of instructions in the storage device 130 and called and / or executed by the processing device 120 .

[0050] Step 310 , acquiring multiple image sequences of the target at the current time point. Step 310 may be executed by the first acquisition module 210 .

[0051] The target is the object for which changes in vascular plaques need to be determined. For example, the target may include the carotid artery, upper limb artery, lower limb artery, thoracic aorta, abdominal aorta, coronary artery, cerebral blood vessels, etc., or any combination thereof.

[0052] The current time point is the time when the target performs the vascular plaque test. For example, the current time point may be December 31, 2021.

[0053] Image sequences refer to images of varying contrast obtained through different MR scan sequences. These contrasts can include T1 contrast and T2 contrast. Because plaque locations appear differently under different sequences, acquiring multiple image sequences can accurately determine plaque composition, location, size, and other information.

[0054] In some embodiments, the plurality of image sequences may include dark blood images and bright blood images. Figure 6A For black blood images, Figure 6BIt is a bright blood image. Black blood images and bright blood images can be obtained through these two imaging modes. Black blood imaging technology suppresses blood flow signals, so the lumen appears black, thereby more clearly showing the vessel wall information and plaque parts. Combining the two imaging technologies of bright blood and black blood is the best imaging examination mode for identifying arterial stenosis and determining the type of plaque. For example, the T1 sequence is black blood imaging, and the lumen appears black and low signal, thereby more clearly showing the vessel wall information and plaque parts, while the TOF sequence is bright blood imaging, and the lumen appears high signal, which can well show occluded blood vessels. Therefore, combining the two magnetic resonance angiography sequences of T1 and TOF can obtain more accurate analysis results.

[0055] In some embodiments, the plurality of image sequences include a T1 sequence, a T1CE sequence, and a TOF sequence; or the plurality of image sequences include a T1 sequence, a T1CE sequence, and a CEMRA sequence. The black blood image includes a T1 sequence and a T1CE sequence, and the bright blood image includes a TOF sequence and a CEMRA sequence.

[0056] The T1 sequence is a longitudinal relaxation time T1-weighted sequence. T1 images can show various cross-sectional anatomical images. The T1CE sequence is a T1 sequence that is performed after injection of contrast agent, which can enhance the vessel wall and / or plaque, facilitating comparative analysis of lesion characteristics. The TOF sequence is a time-of-flight sequence. The CEMRA sequence is an MRA sequence that is performed after injection of contrast agent, which can enhance the display effect. The CEMRA sequence is excellent for displaying large vessel lesions. It can not only clearly show the lesions in the vessels themselves, the relationship between the lesions and the vessels, and the relationship between the vessels and surrounding tissues, but also has excellent display of vascular collateral circulation, with a high positive rate.

[0057] In some embodiments, during MRI imaging, multiple image sequences of the target at the current time point can be acquired by changing the influencing factors of the MR signal.

[0058] Step 320 , acquiring multiple image sequences of the target at one or more historical time points. Step 320 may be executed by the second acquisition module 220 .

[0059] A historical time point is the time when the target underwent vascular plaque testing before the current time point. For example, if the current time point is December 31, 2021, the historical time points can be January 1, 2021, March 31, 2021, June 30, 2021, and September 30, 2021.

[0060] In some embodiments, one or more historical examinations can be automatically detected. Based on the patient ID or other unique identification ID of the current patient, the application searches the database to see if any historical data exists. If so, the application prompts the user to select whether to load the examination information. If the user selects to load the examination information, the application automatically searches for other sequences related to the examination and loads them into the application.

[0061] In some embodiments, the user can manually select inspection data of one or more historical time points and load it into the application, thereby selecting multiple image sequences of the required one or more historical time points and the image sequence of the current time point for vascular plaque analysis.

[0062] Step 330 aligns multiple image sequences including the current time point and each of one or more historical time points to obtain an alignment result of the image sequence at each time point. Step 330 may be performed by the alignment module 230 .

[0063] Image alignment refers to the process of matching two or more images. In some embodiments, a target region can be extracted from one type of image sequence. This region can then be used as a reference image to align other types of image sequences. This ensures that the same object corresponds to the same spatial location in different types of image sequences, thereby obtaining the target region in the other types of image sequences. In some embodiments, when the image sequence includes a T1 image sequence, the T1 image sequence is generally used as the reference image.

[0064] Due to the high contrast of magnetic resonance imaging (MRI) for soft tissue, plaque morphology and composition can be clearly visualized in images. Therefore, plaques and their components can be comprehensively identified by combining multiple MRI image sequences. For multiple MRI image sequences from the same patient, the brightness of the same tissue can vary between sequences, and imaging times can differ by minutes or even tens of minutes. The position of the same tissue can also shift between sequences due to factors such as patient movement. Therefore, rigid image registration techniques are needed to align the images from these sequences. Image registration is an iterative optimization process for matching two images, which often have similarities. One image is kept fixed as a reference image, while the other image is subjected to transformations such as rotation, translation, shearing, and scaling to achieve alignment with the reference image. Rigid image registration only involves rotation and translation transformations. Multi-sequence image registration requires a single image sequence as the reference image, with the remaining images being sequentially registered to it.

[0065] In some embodiments, images can be aligned for comparative viewing. For example, when switching between image sequences at the current time point, image sequences at previous time points are updated in tandem, thereby achieving positional linkage between the current and previous image sequences. In some embodiments, images can be aligned for single-check viewing. For example, multiple image sequences at the same time point can be compared and viewed (e.g., T1 and TOF sequences can be viewed simultaneously), and support for positional linkage between multiple image sequences at the same time point.

[0066] Image sequence alignment can bring together several images of the same patient at the same time point and / or different time points for synchronous viewing and quantitative analysis, thereby obtaining comprehensive information about the patient and improving the accuracy of vascular plaque analysis.

[0067] Step 340 , based on the image sequence alignment result, determines the reconstructed image at each time point. Step 340 may be performed by the reconstruction module 240 .

[0068] The reconstructed image is an image obtained by image reconstruction. In some embodiments, the reconstructed image includes a CPR reconstructed image and an MPR reconstructed image.

[0069] The CPR reconstructed image straightens the twisted, shortened, and overlapping vascular structures and displays them on the same plane. In some embodiments, the CPR reconstructed image can show the overall condition of the blood vessels and lumens.

[0070] In some embodiments, the processing device 120 can automatically extract the center lines of multiple image sequences and then reconstruct along the center lines to obtain CPR reconstructed images. Figure 7A and 7B As shown, the processing device 120 can automatically Figure 7B The centerline Z of the blood vessels in the image is extracted, and then reconstructed along this centerline to obtain the CPR reconstructed image 7A. The CPR reconstructed image is a cross-sectional view reconstructed along the blood vessel centerline and can be reconstructed at any angle along the centerline. After extraction, the current time point is the primary focus. When a specific blood vessel at the current time point is selected, CPR images of that vessel at previous time points are automatically displayed, allowing comparison of the overall condition of the blood vessels and lumen at different time points.

[0071] The MPR reconstructed image is an image of a cross section of a blood vessel. In some embodiments, the MPR reconstructed image can display the vessel wall, lumen, and surrounding tissue.

[0072] In some embodiments, the processing device 120 can automatically extract the center lines of multiple image sequences and then reconstruct along the center lines to obtain MPR reconstructed images. Figure 7B and 7C As shown, the processing device 120 can automatically Figure 7B The centerline Z of the blood vessel in the image is extracted. MPR reconstructed image 7C is then performed at points a, b, and c on centerline Z. 710 represents the outer wall of the vessel, 720 represents the inner wall, and the MPR image reconstructed at point b contains plaque. The MPR reconstructed image is a cross-section perpendicular to the vessel centerline and can be reconstructed at any point along the centerline. After extraction, the current time point is used as the primary focus. Selecting a vessel at the current time point automatically displays MPR images of that vessel at previous time points, allowing for comparison of the vessel wall, lumen, and surrounding tissue at different time points.

[0073] CPR reconstructed images straighten twisted, shortened, and overlapping structures like blood vessels and colons, presenting them on a single plane. This makes them suitable for observing the overall condition of blood vessels and cavities. MPR reconstructed images, which display cross-sections of blood vessels, are suitable for image analysis of vascular plaques. Combining these two methods allows for more accurate analysis of vascular plaque conditions.

[0074] Step 350 , performing vascular plaque analysis based on the reconstructed image, determining newly added plaques and / or historical plaques, and performing historical plaque analysis on the historical plaques. Step 350 may be performed by the analysis module 250 .

[0075] Vascular plaque analysis is a quantitative analysis of vascular plaque. In some embodiments, vascular plaque analysis can be used to determine newly added plaques and / or historical plaques. In some embodiments, vascular plaque analysis can include plaque detection and stenosis detection.

[0076] New patches are patches that exist at the current time point but not at any previous time point. Historical patches are patches that exist at both the current time point and at any previous time point. For example, if a patch is detected at the current time point, December 31, 2021, but does not exist at the previous time point, September 30, 2021, then it is a new patch. For another example, if a patch is detected at the current time point, December 31, 2021, but also exists at the previous time point, September 30, 2021, then it is a historical patch.

[0077] In some embodiments, the processing device 120 can automatically perform vascular plaque matching. For example, if stenosis or plaque is detected at the current time point, it is automatically associated with the image sequence at the historical time point. If the stenosis or plaque is not present in the image sequence at the historical time point, it is defined as a newly added plaque. If the stenosis or plaque is present in the image sequence at the historical time point, it is defined as a historical plaque, and historical plaque analysis is performed on the historical plaque. This historical plaque analysis includes analysis of changes in plaque size, changes in plaque composition, and / or changes in stenosis degree.

[0078] In some embodiments, the processing device 120 can analyze the vascular cavity, vascular wall, internal morphology and plaque composition (for example, lipid core, fibrous cap, calcification, bleeding or hematoma formation and various inflammatory factors, etc.) of historical plaques to obtain numerical values ​​such as the degree of stenosis, wall thickness, standardized wall index, eccentricity index, plaque size, and plaque composition.

[0079] In some embodiments, the processing device 120 can automatically perform stenosis detection and / or plaque detection for each time point of a historical plaque. Upon detection of a plaque, it can automatically perform stenosis analysis and lumen wall analysis at the location of the plaque. In some embodiments, the processing device 120 can automatically perform composition detection for each time point of a historical plaque, analyzing the composition of the plaque based on the varying brightness of each plaque component across multiple image sequences. Different plaque compositions can lead to different plaque properties. For example, if a plaque has a high concentration of hemorrhagic components, it is considered a vulnerable plaque and relatively dangerous.

[0080] In some embodiments, plaque analysis may also include analysis of MIP images.

[0081] The MIP image is a visualized three-dimensional image obtained through maximum intensity projection technology. In some embodiments, vascular plaque analysis can be performed using MIP images at current time points and historical time points.

[0082] Since MIP images have better vascular continuity, by analyzing MIP images at the current time point and historical time points, it is more convenient to perform an overall comparative analysis of vascular morphology.

[0083] Step 360 , determining changes in vascular plaque based on historical plaque analysis, can be performed by the determination module 260 .

[0084] In some embodiments, the vascular plaque change is the change of the vascular plaque over time. In some embodiments, the vascular plaque change may include one or more of a change in plaque size, a change in plaque composition, and a change in stenosis degree.

[0085] In some embodiments, the processing device 120 may automatically analyze historical plaque size changes and output numerical values ​​and / or curves showing changes in plaque size, plaque volume, and / or maximum plaque area over time to display changes in plaque size.

[0086] In some embodiments, the processing device 120 can automatically analyze historical changes in plaque composition and output numerical values ​​and / or curves showing changes in the volume and proportion of each component of the plaque over time to display changes in plaque composition.

[0087] In some embodiments, the processing device 120 can automatically analyze the change in stenosis degree and output a numerical value and / or a curve showing the change in stenosis degree over time to show the change in stenosis degree. In some embodiments, the relative stenosis degree can be automatically calculated based on the relative stenosis ratio.

[0088] By analyzing the vascular plaque images, the changes in the vascular plaque can be determined, thereby more accurately judging the progression of the plaque.

[0089] Figure 4 is a schematic diagram of a machine learning model 400 according to some embodiments of this specification.

[0090] The future time point is the time after the current time point when the target performs vascular plaque detection. For example, if the current time point is December 31, 2021, the future time point may be March 31, 2022.

[0091] The machine learning model refers to a processing module that can predict changes in vascular plaques to output the vascular plaque status at a future time point. In some embodiments, the machine learning model can be obtained based on the training of a deep learning neural network. For instructions on training the machine learning model, see Figure 5 .

[0092] In some embodiments, the machine learning model can be constructed based on a deep learning neural network model. Exemplary deep learning neural network models may include a convolutional machine learning model (CNN), a fully convolutional neural network (FCN) model, a generative adversarial network (GAN), a back propagation (BP) machine learning model, a radial basis function (RBF) machine learning model, a deep belief network (DBN), an Elman machine learning model, or the like, or a combination thereof.

[0093] In some embodiments, the input to the machine learning model includes changes in vascular plaque. The changes in vascular plaque can be changes in vascular plaque at the current time point and at one or more historical time points. In some embodiments, the changes in vascular plaque can be input into the machine learning model in various feasible ways, such as by combining various parameters of the changes in vascular plaque into a single input feature.

[0094] In some embodiments, the output of the machine learning model includes a predicted vascular plaque condition at a future time point. In some embodiments, the vascular plaque condition at a future time point is obtained by predicting vascular plaque changes based on the trained machine learning model.

[0095] In some embodiments, the processing device 120 can input various parameters of the vascular plaque change into a trained machine learning model, and the machine learning model can output the vascular plaque condition at a future time point based on the various parameters of the vascular plaque change.

[0096] By predicting changes in vascular plaque, the system predicts the condition of vascular plaque at future time points, improving the accuracy of the prediction. When predicting vascular plaque conditions at future time points, the model analyzes changes in vascular plaque at the current time point and at one or more historical time points to determine the condition of vascular plaque at the future time point, significantly improving the accuracy of the predicted vascular plaque condition at the future time point.

[0097] Figure 5 is a schematic diagram of machine learning model training 500 according to some embodiments of this specification.

[0098] In some embodiments, the parameters of the machine learning model can be obtained by training multiple labeled training samples. In some embodiments, multiple sets of training samples 540 can be obtained, each set of training samples can include multiple training data and labels corresponding to the training data. The training data can include changes in vascular plaques at the current time point and one or more historical time points. The labels of the training data can be the vascular plaque conditions at future time points after the changes in vascular plaques at the current time point and one or more historical time points are predicted. The parameters of the initial machine training model 550 can be updated using multiple sets of training samples 540 to obtain a trained initial machine learning model 550. The parameters of the machine learning model 520 are derived from the trained initial machine learning model 550. The parameters can be transferred in any common manner.

[0099] In some embodiments, the training data and labels of the machine learning model can be obtained from historical data. For example, the training data can be obtained from historical image data collected by imaging equipment during historical medical procedures. In some embodiments, the training data and labels of the machine learning model can also be obtained through manual input, calling relevant interfaces, etc. In some embodiments, any other method can be used to obtain training samples and labels of training samples for the machine learning model.

[0100] In some embodiments, the labels for the training data can be obtained based on historical data. For example, the historical data contains vascular plaque conditions at N+1 time points, where the vascular plaque conditions at time points 1, 2, 3, ..., N-1 are historical time points; the vascular plaque conditions at time point N are current time points; and the vascular plaque conditions at time point N+1 are future time points. The vascular plaque changes at time points 1, 2, 3, ..., N serve as training data, and the vascular plaque condition at time point N+1 serves as the label for the training data.

[0101] In some embodiments, the parameters of the initial machine learning model 550 can be iteratively updated based on multiple training samples so that the loss function of the model meets preset conditions. For example, the loss function converges, or the loss function value is less than a preset value. When the loss function meets the preset conditions, the model training is completed, and the trained initial machine learning model 550 is obtained. Among them, the machine learning model 520 and the trained initial machine learning model 550 have the same model structure. Specifically, the input of the trained initial machine learning model 550 is each training sample, and the output is the vascular plaque situation at a future time point corresponding to each training sample. Correspondingly, the input of the machine learning model 520 is the vascular plaque change situation 510, and the output is the vascular plaque situation 530 at a future time point.

[0102] By training the machine learning model, the trained machine learning model has a predictive function. Using this function, we can predict the changes in vascular plaques at the current time point and historical time points, predict the vascular plaque conditions at future time points, and improve the accuracy of the prediction.

[0103] In some embodiments, an image analysis device includes a processor and a memory; the memory is used to store instructions, and when the instructions are executed by the processor, the device implements the image analysis method.

[0104] In some embodiments, a computer-readable storage medium stores computer instructions. When a computer reads the computer instructions in the storage medium, the computer executes the image analysis method.

[0105] In some embodiments of this specification, by aligning, reconstructing, and analyzing image sequences at the current time point and historical time points, the changes in vascular plaque can be quantitatively determined, resulting in a more accurate picture of vascular plaque changes. Furthermore, by using machine learning models to predict vascular plaque conditions at future time points, a better understanding of the future development of vascular plaque can be achieved.

[0106] While the basic concepts have been described above, it will be apparent to those skilled in the art that the detailed disclosure is merely illustrative and does not limit this specification. Although not explicitly stated herein, various modifications, improvements, and revisions to this specification may be made by those skilled in the art. Such modifications, improvements, and revisions are suggested in this specification and remain within the spirit and scope of the exemplary embodiments of this specification.

[0107] This specification also uses specific terms to describe the embodiments of this specification. For example, "one embodiment," "an embodiment," and / or "some embodiments" refer to a feature, structure, or characteristic associated with at least one embodiment of this specification. Therefore, it should be emphasized and noted that references to "one embodiment," "an embodiment," or "an alternative embodiment" two or more times in different locations in this specification do not necessarily refer to the same embodiment. Furthermore, certain features, structures, or characteristics of one or more embodiments of this specification may be appropriately combined.

[0108] In addition, unless expressly stated in the claims, the order of the processing elements and sequences, the use of alphanumeric characters, or the use of other names described in this specification are not intended to limit the order of the processes and methods of this specification. Although the above disclosure discusses some of the invention embodiments currently considered useful through various examples, it should be understood that such details are for illustrative purposes only, and the appended claims are not limited to the disclosed embodiments. On the contrary, the claims are intended to cover all modifications and equivalent combinations that are consistent with the spirit and scope of the embodiments of this specification. For example, although the system components described above can be implemented by hardware devices, they can also be implemented only by software solutions, such as installing the described system on an existing server or mobile device.

[0109] Similarly, it should be noted that, in order to simplify the presentation of this specification and thus facilitate understanding of one or more embodiments of the invention, the foregoing descriptions of the embodiments of this specification sometimes combine multiple features into a single embodiment, figure, or description thereof. However, this disclosure method does not imply that the subject matter of this specification requires more features than those recited in the claims. In fact, an embodiment may have fewer features than all of the features of a single disclosed embodiment.

[0110] In some embodiments, numbers are used to describe the quantity of components and attributes. It should be understood that such numbers used in the description of the embodiments are modified by the modifiers "about", "approximately" or "substantially" in some examples. Unless otherwise stated, "about", "approximately" or "substantially" indicate that the numbers are allowed to vary by ±20%. Accordingly, in some embodiments, the numerical parameters used in the description and claims are approximate values, which may change according to the required characteristics of individual embodiments. In some embodiments, the numerical parameters should take into account the specified significant digits and adopt the general method of retaining digits. Although the numerical domains and parameters used to confirm the breadth of their range in some embodiments of this specification are approximate values, in specific embodiments, the settings of such numerical values ​​are as accurate as possible within the feasible range.

[0111] Each patent, patent application, patent application publication, and other materials, such as articles, books, specifications, publications, and documents, cited in this specification is hereby incorporated by reference in its entirety. This includes application history documents that are inconsistent with or conflict with the content of this specification, as well as documents (currently or subsequently attached to this specification) that limit the broadest scope of the claims of this specification. It should be noted that if the descriptions, definitions, and / or terminology used in the accompanying materials are inconsistent or conflicting with the content of this specification, the descriptions, definitions, and / or terminology used in this specification will control.

[0112] Finally, it should be understood that the embodiments described in this specification are intended only to illustrate the principles of the embodiments of this specification. Other variations may also fall within the scope of this specification. Therefore, by way of example and not limitation, alternative configurations of the embodiments of this specification may be considered consistent with the teachings of this specification. Accordingly, the embodiments of this specification are not limited to the embodiments explicitly described and illustrated in this specification.

Claims

1. A method for image analysis, characterized in that: include: Acquiring multiple image sequences of the target at a current time point, the multiple image sequences being images obtained through different sequences of MR scans, and the multiple image sequences including black blood images and bright blood images; Acquiring a plurality of image sequences of the target at one or more historical time points; Aligning the multiple image sequences including the current time point and each time point of the one or more historical time points, and obtaining an alignment result of the image sequence at each time point; Determining a reconstructed image at each time point based on the image sequence alignment result; performing vascular plaque analysis based on the reconstructed image, determining newly added plaques and historical plaques, and performing historical plaque analysis on the historical plaques; and The change of vascular plaque is determined based on the historical plaque analysis.

2. The method according to claim 1, characterized in that The changes in the vascular plaque include one or more of changes in plaque size, changes in plaque composition, and changes in the degree of stenosis.

3. The method according to claim 1, characterized in that Aligning the multiple image sequences including the current time point and each time point of the one or more historical time points, and obtaining an alignment result of the image sequence at each time point includes: The plurality of image sequences at each time point are aligned using a rigid image registration technology to obtain an alignment result of the image sequence at each time point.

4. The method according to claim 1, wherein The reconstructed image includes a CPR reconstructed image and an MPR reconstructed image.

5. The method according to claim 1, wherein The black blood image includes a T1 sequence and a T1CE sequence, and the bright blood image includes a TOF sequence; or, the black blood image includes a T1 sequence and a T1CE sequence, and the bright blood image includes a CEMRA sequence.

6. The method according to claim 1, characterized in that The method further comprises: Based on the changes in the vascular plaques, the vascular plaque conditions at future time points are predicted using a machine learning model.

7. An image analysis system, characterized in that: include: a first acquisition module, configured to acquire a plurality of image sequences of the target at a current time point, wherein the plurality of image sequences are images obtained by different sequences of MR scans, and the plurality of image sequences include black blood images and bright blood images; A second acquisition module is used to acquire a plurality of image sequences of the target at one or more historical time points; an alignment module, configured to align the plurality of image sequences including the current time point and each of the one or more historical time points, and obtain an alignment result of the image sequence at each time point; A reconstruction module, configured to determine a reconstructed image at each time point based on the image sequence alignment result; an analysis module, configured to perform vascular plaque analysis based on the reconstructed image, determine newly added plaques and historical plaques, and perform historical plaque analysis on the historical plaques; as well as A determination module is used to determine changes in vascular plaques based on the historical plaque analysis.

8. The system according to claim 7, characterized in that The changes in the vascular plaque include one or more of changes in plaque size, changes in plaque composition, and changes in the degree of stenosis.

9. An image analysis device, comprising a processor and a memory; the memory is used to store instructions, and when the instructions are executed by the processor, the device implements the image analysis method according to any one of claims 1 to 6.

10. A computer-readable storage medium storing computer instructions. When a computer reads the computer instructions in the storage medium, the computer executes the image analysis method according to any one of claims 1 to 6.

Images