Methods and systems for determining vascular region pathology based on radiodensity and radiomic features

Abstract

The invention relates to methods for characterizing the pathology or status of a vascular region of a subject, by: providing medical imaging data acquired along the length of a blood vessel, and: determining radial segments of the blood vessel extending from a point in the center of the blood vessel to a distance extending beyond the outer wall of the blood vessel to the surrounding region of the blood vessel; determining values of radiodensity values and / or one or more radiomic features of the radial segments; and comparing or using the absolute values of each of the determined values to predetermined values to generate an output value indicative of the pathology or status of the vascular region or plaque inflammation. At least one of the method steps is computer implemented. The invention also provides computer program products configured to perform these methods, and systems comprising one or more processors configured to perform these methods.

Description

Technical Field

[0001] This invention relates to methods for characterizing the pathology or condition of a vascular region in a subject, and methods for characterizing plaque in a subject's blood vessels, wherein at least one method step is implemented by a computer. The methods can be used to monitor plaque progression, predict the risk of a subject experiencing cardiovascular events, guide drug treatment decisions, and monitor responses to medical care.

[0002] The present invention also provides a computer program product configured to perform these methods, and a system comprising one or more processors. Background Technology

[0003] Atherosclerosis is a progressive process in which the arterial walls thicken due to the invasion and accumulation of white blood cells. This inflammatory process leads to the formation of plaques within the blood vessel walls, containing live white blood cells, dead cell debris, and fatty deposits including cholesterol and triglycerides.

[0004] Stable atherosclerotic plaques are usually asymptomatic and are typically rich in extracellular matrix and smooth muscle cells, while unstable plaques are rich in macrophages and foam cells. In unstable plaques, the extracellular matrix separating the lesion from the arterial lumen (also known as the fibrous cap) is often thin and prone to rupture. Rupture of the fibrous cap can eventually induce blood clot formation within the lumen. Such clots can block the artery or break off, enter the circulation, and eventually block smaller downstream vessels, leading to thromboembolism. Chronic dilated plaques are often asymptomatic until the vascular occlusion (stenosis) becomes severe enough to cause insufficient blood supply to downstream tissues.

[0005] Atherosclerosis often goes unnoticed for decades because arteries can dilate at the site of plaque without immediately affecting blood flow. In fact, plaque rupture is also asymptomatic unless it causes the artery to narrow or close sufficiently, thus obstructing blood flow to different organs and triggering symptoms. Typically, the disease is only diagnosed when a patient experiences another cardiovascular event, such as a stroke or heart attack.

[0006] Atherosclerosis can cause narrowing of the coronary arteries, which are responsible for delivering oxygenated blood to the heart. This can cause symptoms such as angina, chest pain, shortness of breath, sweating, nausea, dizziness or lightheadedness, difficulty breathing, or palpitations. Cardiac arrhythmias can also be caused by ischemia of the heart. Atherosclerosis that narrows the carotid arteries supplying the brain and neck can cause symptoms such as weakness, difficulty thinking clearly, difficulty speaking, dizziness and difficulty walking or standing, blurred vision, numbness in the face, arms, and legs, severe headaches, and loss of consciousness. These symptoms can also occur in a stroke caused by significant narrowing or blockage of the arteries supplying the brain, leading to cerebral ischemia and brain cell death. Peripheral arteries supplying the legs, arms, and pelvis can also be affected. Symptoms may include numbness and pain in the affected limbs. Plaque formation can also occur in the renal arteries supplying the kidneys. The occurrence and accumulation of plaque leads to reduced blood flow to the kidneys and chronic kidney disease, which, like all other sites, is usually asymptomatic until the late stages.

[0007] Vascular inflammation is a key feature of atherosclerosis formation and plays a crucial role in atherosclerotic plaque stability by triggering plaque rupture that leads to acute coronary syndrome (see Ross R. N EN). g (J Med 1999;340:115-26, and Major AS et al. Circulation 2011;124:2809-11). Importantly, more than 50% of acute coronary syndromes are caused by highly inflammatory but anatomically insignificant atherosclerotic plaques (Fishbein MC et al. Circulation 1996;94:2662-6).

[0008] Cardiac computed tomography angiography (CCTA) can quantify the extent, distribution, and characteristics of coronary artery plaques, but this is insufficient for optimal risk prediction for individuals. It is well known that most acute myocardial infarctions (MIs) are secondary to vascular occlusion with mild coronary plaque disease (erosion or rupture). This is related to the biological characteristics of the underlying coronary plaques, particularly inflammation.

[0009] In the PROMISE trial, 54% of adverse events occurred in patients without significant stenosis, compared to only 12% of those receiving CCTA. Therefore, in the CCTA population, more than half of the total risk of adverse cardiovascular events goes unrecognized by coronary artery stenosis. This limitation is a driving factor contributing to the “residual risk” of adverse cardiovascular outcomes despite efforts to manage cardiovascular disease (CVD) according to current recommendations.

[0010] CCTA can identify patients with high-risk plaque-related features such as low-attenuation plaques, napkin ring sign, positive remodeling, and punctate plaque calcification; however, these provide only modest incremental information in individual patients. The predictive value of high-risk plaque (HRP) features has been investigated in the PROMISE and SCOT-HEART trials. In the PROMISE study, approximately 15% of patients were found to have HRP on their CCTA, and HRP was associated with more adverse cardiovascular events, although the major adverse cardiovascular event (MACE) endpoint includes revascularization, which may not reflect the additional value of HRP beyond stenosis severity.

[0011] Nevertheless, most patients with HRP did not experience cardiovascular events, while many without HRP did, suggesting that the predictive value of HRP on CCTA is limited. In fact, of the 1019 HRPs identified on CCTA, only 24 subsequently resulted in nonfatal MI, indicating that the absolute risk of a single plaque-related cardiovascular event identified at a single time point is extremely low.

[0012] This observation is consistent with other CCTA and intravascular ultrasound (IVUS) studies. In the SCOT-HEART trial, HRP features were detected on 1376 CCTAs in 608 out of 1769 participants. Subjects with HRP features had an increased likelihood of adverse cardiac events during follow-up, but the absolute increase in risk was very small (4.1% with HRP vs. 1.4% without HRP). Importantly, more than one-third of the events occurred in subjects without HRP features. More detailed analysis identified low-attenuation non-calcified plaque burden as the most specific HRP feature for predicting adverse events. Furthermore, these recent findings highlight the importance of cellular inflammation mechanisms in the vessel wall as a driver of disease progression and event risk.

[0013] In addition to contributions from stenosis, flow restriction, or high-risk plaque characteristics, coronary inflammation is a major factor in the progression of CAD and a key determinant of high-risk plaques that drive adverse clinical events.

[0014] Imaging perivascular adipose tissue (PVAT) around the coronary arteries (pericardial adipose tissue; PCAT) has become a promising technique for imaging inflammation in the coronary artery wall. A key recent finding is that PVAT “senses” the presence of inflammation in the coronary artery wall. These signals transduce changes in PVAT differentiation, leading to smaller, less lipid-rich adipocytes, larger inflammatory cell infiltration, and higher tissue water content. These changes alter tissue attenuation values ​​in the 3D distribution around the coronary arteries, which can be detected using cardiac computed tomography angiography (CCTA). Simple measurements of PVAT attenuation require correction for anatomical, technical factors, as well as patient and clinical variables. However, the effectiveness of measuring changes in PCAT attenuation has been demonstrated in studies using similar techniques across different patient populations.

[0015] Goeller et al. (JAMA Cardiol. 2018; 3:858) reported increased uncorrected PCAT CT attenuation around culprit lesions compared to non-culprit lesions and matched controls in ACS patients. In multivariate analysis, low- and moderate-attenuation non-calcified plaque burden and PCAT CT attenuation were independently associated with the presence of culprit lesions. They concluded that combining quantitative high-risk plaque features with PCAT CT attenuation can more reliably identify vulnerable plaques.

[0016] The inventors' recent ORFAN (Oxford Risk Factors and Noninvasive Imaging) study validated the effectiveness of the Fat Fat Attenuation Index (FAI) score and assessed its real-world impact on patient management. This study revealed that cardiac death or MACE occurred in 3.4% and 8.7% of patients without obstructive CAD, respectively. In patients with obstructive or nonobstructive disease, those with a FAI score above the 75th percentile in LAD had a 20.2-fold and 6.7-fold higher risk of cardiac death or MACE, respectively, compared to the 25th percentile. In patients without obstructive CAD, the risk of cardiac death or MACE remained 10.5-fold and 4.8-fold higher, respectively. An AI-assisted prognostic model (AI-risk), incorporating inflammatory risk based on the FAI score, significantly reclassified patients' clinical risk across the entire cohort and in patients without obstructive CAD compared to a clinical risk factor-based predictive model (QRISK3), leading to treatment changes in a significant proportion of patients, such as statin initiation, statin dose escalation, and / or additional treatments, such as colchicine. The study concluded that patients with high coronary artery inflammation, as measured by the FAI score, had a significantly higher risk of developing MACE, and that AI-Risk led to a substantial proportion of patients undergoing routine CCTA reclassification and changes in treatment management. Therefore, this CCTA risk score can be used as a tool for precision medicine.

[0017] While such CCTA methods have proven promising in the monitoring and treatment of coronary artery disease in recent years, there is still a need to develop more precise and reliable methods to identify vulnerable vascular regions and plaques. In particular, there is a need to detect vascular inflammation and the development of high-risk fatty plaques earlier and to minimize false negatives obtained using previously known but less sensitive functional biomarkers for CCTA. Summary of the Invention

[0018] In a first aspect, the present invention provides a method for characterizing the pathology or condition of a vascular region in a subject, the method comprising:

[0019] (a) Provide medical imaging data acquired along the length of the blood vessel; and

[0020] (i) Determine the radial segment of the blood vessel extending from its centerline to a distance extending beyond the outer wall of the blood vessel to the area surrounding the blood vessel;

[0021] (ii) Determine the values ​​of the radiodensity and / or one or more radiomics characteristics of the radial segment; and

[0022] (b) Compare each value determined in (ii) with a predetermined value, or use an absolute value, to generate an output value indicating the pathology or condition of the vascular region.

[0023] At least one method step is implemented by a computer.

[0024] According to a second aspect, the present invention provides a method for characterizing plaques in the blood vessels of a subject, the method comprising:

[0025] (a) Providing medical imaging data acquired along the length of the blood vessel; and

[0026] (i) Identify voxels corresponding to plaques by their radioactivity;

[0027] (ii) Determine radial segments of the blood vessel containing voxels corresponding to the plaque, wherein the radial segments extend from the centerline of the blood vessel to a distance extending beyond the outer wall of the blood vessel;

[0028] (iii) Determine the values ​​of the radioactivity density and / or one or more radiomics characteristics of the radial segment;

[0029] (b) Compare each value determined in (iii) with a predetermined threshold, or use the absolute value determined in (iii) to generate an output value indicating inflammation of the plaque.

[0030] At least one method step is implemented by a computer.

[0031] The methods described in the first and second aspects can be used to monitor plaque progression, predict the risk of cardiovascular events in subjects, guide drug treatment decisions, monitor medical responses by assessing dynamic changes in coronary artery inflammation, non-invasively monitor aortic aneurysms or carotid plaques, stratify subjects based on their risk of cardiac death or cardiovascular events, quantify vascular inflammation, fibrosis, edema, and vascular distribution, and calculate local changes in inflammatory status.

[0032] Preferably, the output values ​​are corrected for biological and technical factors to improve accuracy and sensitivity. Preferably, these corrections are performed by a processing system (i.e., by a computer).

[0033] In a third aspect, the present invention provides a computer program product including executable instructions that, when executed by one or more processors, cause the one or more processors to perform the method of the first aspect or the second aspect.

[0034] In a fourth aspect, a system is provided, comprising one or more processors configured to perform the methods of the first or second aspect. Attached Figure Description

[0035] Figure 1A A blood vessel is depicted having a lumen 100, a centerline 101, a vessel wall 102, and radial segments extending from the centerline 101 to a predetermined distance 104 in the space surrounding the vessel 103. Figure 1B The illustration shows the same blood vessel, but with plaque 105 in the vessel wall. The radial segments in this example are defined by the physical limits of the plaque.

[0036] Figure 2A and 2B Similarly, a blood vessel is illustrated, having a lumen 200, a central line 201, a vessel wall 202, and radial segments extending from the central line 201 to a predetermined distance 204 in the space surrounding the vessel 203. Figure 2A The thickness of the radial segment extends along the length of the vessel, while Figure 2B The thickness of the mid-radial segment is only a portion of the length of the blood vessel.

[0037] Figure 3 The illustration depicts a blood vessel having a lumen 300, a centerline 301, a vessel wall 302, and a perivascular space 303 extending a predetermined distance 304 from the vessel wall 302. The vessel is divided into multiple radial segments separated by dashed lines, each having a predefined geometry. Patches 305 exist within the vessel wall 302, spanning the multiple radial segments of the predefined geometry 304.

[0038] Figure 4A system example suitable for implementing the embodiments of the method is schematically illustrated. System 1100 includes at least one server 1110 that communicates with a reference data repository 1120. The server may also communicate with other hardware or systems 1130 operated by healthcare professionals, for example, via a communication network 1140.

[0039] Figure 5 An example of a suitable server 1110 that can be used to implement the methods of the present invention is described. In this example, the server includes at least one microprocessor 1200, a memory 1201, optional input / output devices 1202 (such as a keyboard and / or display), and an external interface 1203 interconnected via a bus 1204 as shown in the figure. In this example, the external interface 1203 can be used to connect the server 1110 to peripheral devices, such as a communication network 1140, a reference data repository 1120, other storage devices, etc.

[0040] Figure 6 A digital phantom of the coronary artery constructed in Example 1 is shown. Image a) shows a non-inflammatory coronary artery, b) shows a coronary artery with added focal inflammatory areas, c) shows the area where PVAT attenuation was measured according to conventional methods (i.e., the entire circumference of the vessel was analyzed and scored), and d) shows the area where PVAT attenuation was measured within a quarter radial segment of the PVAT region around the coronary artery according to the present invention.

[0041] Figure 7 The increased inflammation of the digital phantom of the coronary artery, as measured according to Example 1, is shown, with the inflammation intensity increasing from -75 HU to -49 HU.

[0042] Figure 8 The results of 50 simulations of the PVAT attenuation of a digital phantom of the coronary artery at each inflammatory intensity are shown in Example 1, comparing the full PVAT region (i.e., the entire circumference) with a quarter PVAT region (i.e., the radial segment according to the invention). Figure 8 The horizontal dashed line in the chart shows the upper quarter distance of the baseline using the radial segment method, while the horizontal dotted line shows the equivalent upper quarter distance using the full-circle method.

[0043] Figure 9 The results of PVAT attenuation in 500 simulations of a digital phantom of the coronary artery at each inflammatory intensity in Example 1 are shown. The simulations used the full PVAT region (i.e., the entire circumference) and a quarter PVAT region (i.e., the radial segment according to the invention). Detailed Implementation

[0044] The inventors have developed computer-implemented methods capable of characterizing the pathology or condition (e.g., its phenotype, such as composition and / or texture) of vascular regions in subjects from medical imaging data with greater sensitivity and accuracy. Therefore, these methods mitigate false-negative results frequently obtained from known analytical methods that lack the sensitivity to identify small, highly inflammatory vascular regions or plaques.

[0045] Characterization of vascular regions and / or plaques involves determining output values ​​based on radioactivity and / or one or more radiomics features of radial segments of the vessel (e.g., wedge-shaped portions). These radial segments extend from the vessel's centerline (which can be automatically determined by the processing system, i.e., computer-implemented) to a predetermined distance from the vessel's outer wall. This analysis can identify inflammatory vascular regions around the vessel and characterize plaques associated with these inflammatory regions based on a more limited, localized analysis of the target vessel. Focusing the analysis on specific radial segments allows for more sensitive analysis and improved clarity of detail within the vascular region, thus aiding in the detection of high-risk features and abnormalities that might be overlooked using current methods (analyzing vascular pathology or condition by assessing the entire vessel and / or the surrounding space).

[0046] This targeted approach leads to more accurate and precise diagnoses, enabling clinicians to make more informed decisions regarding patient care and treatment plans. Furthermore, radial segments provide smaller datasets that can be visualized more effectively, allowing clinicians to more easily and specifically navigate medical imaging data, ultimately facilitating communication in discovering and developing targeted treatment plans.

[0047] The method of the present invention can be used to provide clinically valuable outputs that are capable of: monitoring plaque progression, predicting the risk of a subject to cardiovascular events, guiding drug treatment decisions, monitoring responses to medical care, non-invasively monitoring aortic aneurysms or carotid plaques, and stratifying subjects based on their risk of cardiac death or cardiovascular events; quantifying one or more of the following: vascular inflammation, fibrosis, edema, and vascular distribution; and calculating local changes in inflammation or other pathological conditions.

[0048] Subjects may be individuals diagnosed with a vascular inflammation-related condition, or individuals suspected of having or at risk of having a vascular inflammation-related condition, particularly vascular inflammation affecting the coronary arteries. Alternatively, patients may be healthy individuals who have not been diagnosed with a vascular inflammation-related condition and / or whose risk of having a vascular inflammation-related condition is unknown.

[0049] By limiting the analysis of vascular pathology or condition to local radial segments, rather than analyzing the entire vessel and / or perivascular space, small but aggressive areas of inflammation and / or high-risk plaques can be detected. This is an improvement over known methods that analyze and quantify the average vascular and / or perivascular tissue across the entire vessel or perivascular space (including or around the entire circumference of the vessel). Therefore, the method of the present invention can identify inflammatory areas with higher sensitivity and an improved signal-to-noise ratio than previously known methods, which in turn reduces false positives and enables earlier detection and intervention.

[0050] The vascular region and / or plaque characterization of the present invention can be used as an adjunct tool in routine clinical CT angiography to identify patients at high risk of cardiac events and mortality, including as a sensitive and specific screening tool in populations that appear healthy and low-risk based on conventional interpretation of their scans. Therefore, the claimed method has applications in both primary prevention (healthy individuals not yet diagnosed with heart disease) and secondary prevention (patients diagnosed with coronary artery disease) to identify individual risk profiles beyond conventional risk factors, guide pharmacological treatment decisions, and monitor responses to appropriate medical care.

[0051] Furthermore, the current method also allows for mapping of high-risk vascular regions and / or high-risk plaques, which helps to monitor disease progression more effectively, improve screening efficiency, and provide more targeted care.

[0052] In a first aspect, the present invention provides a method for characterizing the pathology or condition of a vascular region in a subject, the method comprising:

[0053] (a) Provide medical imaging data acquired along the length of the blood vessel; and

[0054] (i) Determine the radial segment of the blood vessel extending from its centerline to a distance extending beyond the outer wall of the blood vessel to the area surrounding the blood vessel;

[0055] (ii) Determine the values ​​of the radiodensity and / or one or more radiomics characteristics of the radial segment; and

[0056] (b) Compare each value determined in (ii) with a predetermined value, or use an absolute value, to generate an output value indicating the pathology or condition of the vascular region;

[0057] At least one of the method steps is implemented by a computer.

[0058] Determine radial segment

[0059] According to each aspect of the invention, the radial segment is a wedge shape in the circumferential plane of the blood vessel, extending from a point at the center of the blood vessel to a distance extending beyond the outer wall of the blood vessel to the area surrounding the blood vessel.

[0060] This can be used as a reference. Figure 1A and 1B To illustrate, the diagram schematically depicts radial segments drawn with dashed lines. As can be seen from these figures, the wedge-shaped radial segments extend radially from a point 101 at the center of the vessel within the vessel lumen 100 to the space surrounding the vessel 103, reaching a predetermined distance 104 from the outer wall 102 of the vessel. Vessel portions not included in the radial segments are not used to calculate values ​​for the vessel's radioactivity and / or one or more radiomics characteristics.

[0061] Radial segments can have predefined geometries. For example, it can be a portion of a blood vessel cross-section in the radial direction.

[0062] The angle of the radial segment from a point in the center of the blood vessel can be 1 to 359°, preferably 2 to 270°, more preferably 10 to 180°, and even more preferably 30 to 150°. This can be referenced. Figure 3 The visualization illustrates a blood vessel having a lumen 300, a center point 301, a vessel wall 302, and a perivascular space 303 extending a predetermined distance 304 from the vessel wall 302. The vessel is divided into multiple radial segments separated by dashed lines, each with a predefined geometry—in this case, each radial segment is 1 / 8 of the vessel and has an angle of 45°. Patches 305 are present in the vessel wall 302, spanning the multiple radial segments with predefined geometries.

[0063] The method may include determining a point 301 in the center of a blood vessel from medical imaging data, determining radial segments of the blood vessel and analyzing each segment in sequence to characterize the pathology or condition of each vascular region, or to characterize any plaques 305 that may be present in the vessel wall 302.

[0064] The term "point in the center of a blood vessel" refers to a point in medical imaging data that represents a point within the lumen. This can be the center point of the blood vessel or another point within the lumen. Preferably, a point in the center of a blood vessel corresponds to the centerline of the blood vessel. Methods for determining the centerline are well known to those skilled in the art.

[0065] Biological factors can also influence the geometry of radial segments. For example, if a potentially high-risk or inflammatory plaque can be identified from medical imaging data, the segment can be adjusted to contain voxels corresponding to the entire plaque or a portion of it. For instance, if the plaque is located on a quarter of a blood vessel, the radial segment could be a quarter of the blood vessel and would contain voxels corresponding to the plaque. Figure 1B An embodiment is illustrated, wherein the geometry of the radial segments is defined by the physical limits of patch 105.

[0066] The same applies to vascular regions that do not contain plaques but can be identified from medical imaging data as highly inflammatory, edematous, fibrotic, or abnormally vascularized by radiodensity and / or radiomics measurements of the first aspect (ii) or the second aspect (iii) of this invention. In these cases, the radial segments can be adjusted to include voxels corresponding to discrete, inflammatory regions.

[0067] The thickness of the radial segments can vary. For example, refer to... Figure 2A The radial segment may extend along the length 206 of the analyzed vessel. Alternatively, the radial segment may have only a thickness that is part of the length 206 of the vessel, such as... Figure 2B As shown.

[0068] The minimum thickness of a radial segment is one voxel. However, it is preferable to have a width of multiple voxels to improve the signal-to-noise ratio. Increasing the number of sampled voxels improves the estimation of the true mean of the voxels by reducing the "standard error of the mean" by a ratio of 1 / sqrt(n), where "n" is the number of sampled voxels.

[0069] If plaques or discrete, highly inflammatory areas are present in the blood vessels, the thickness of the radial segments can be adjusted to include voxels corresponding to the plaques or discrete inflammatory areas.

[0070] For a given medical imaging dataset, multiple radial segments can be identified along the length of the blood vessel, which may have varying geometries. They may also have the same geometry, such as having the same wedge angle in a given measurement, like... Figure 2A As shown.

[0071] The distance extending beyond the outer wall of a blood vessel can depend on the type of blood vessel being analyzed and its spatial location relative to other body tissues or other surrounding blood vessels. For example, if a particular blood vessel being analyzed is close to another type of body tissue, the space surrounding the vessel should not contain voxels corresponding to that other body tissue.

[0072] The distance extending beyond the outer wall of the vessel can also be determined by the radiometric density in the image (for the radial range) or based on the patient's anatomy. For example, the perivascular tissue is thinner further distal to the vessel; or to prevent the inclusion of other anatomical features, such as excluding tissue outside the pericardial sac in the case of coronary arteries.

[0073] Preferably, the distance extending beyond the outer wall of the blood vessel is determined by the processing system (i.e., this step is performed by a computer), which preferably calculates this distance based on a reference dataset. Alternatively, this distance can be manually determined or adjusted by the operator of the processing system.

[0074] In one embodiment, the distance extending beyond the outer wall of the blood vessel can be:

[0075] i) A standard distance that is not equal to or related to the baseline vessel diameter or radius;

[0076] ii) is a derived value, multiple of, or equal to the radius or diameter of the basic vessel; or

[0077] iii) 0.1 mm to 3 cm, preferably 0.2 mm to 2.5 cm, more preferably 0.3 mm to 2.25 cm, and even more preferably 0.4 mm to 2 cm.

[0078] The inventors chose a wedge shape for the radial segments rather than other geometries because they believed it could accurately capture the paracrine effects of mediators released from the vessel wall on adjacent perivascular tissues due to the presence of vascular inflammation and oxidative stress (see, for example, Margaritis et al. Circulation 2013; 127(22):2209-21). In addition to mediators released from the vessel wall into the perivascular space, perivascular adipose tissue is thought to “sense” inflammation in adjacent vessel walls, leading to changes in PVAT differentiation. This communication between inflammatory vascular cells and adjacent perivascular tissue is believed to occur radially. Therefore, it is envisioned that radial segments extending beyond the vessel wall in the same direction can provide a picture of the effects of any vascular inflammation, edema, fibrosis, or blood supply status within the same radial segment on the spatial biology of the perivascular space. This specificity would be lost by methods that use different geometries for analysis when “magnifying” medical imaging data.

[0079] In addition to radial lines, another method for defining quantitative segments is to simulate the diffusion of paracrine mediators from the plaque / vessel wall interface through perivascular tissue to identify tissue areas affected by paracrine mediators.

[0080] This can be achieved using reaction-diffusion equations. For example, diffusion equations (Fick's second law) can be used to simulate the diffusion of cytokines or signaling molecules through perivascular tissue.

[0081]

[0082] in:

[0083] c is the concentration of cytokines or signaling molecules at a given point.

[0084] D is the diffusion coefficient (specific to cytokines or signaling molecules).

[0085] It is the Laplace operator. This represents the concentration gradient.

[0086] x, y, and z define the position in three-dimensional space.

[0087] t represents time.

[0088] Reaction steps can be added to the diffusion equation above, for example, to represent the metabolism of cytokines or signaling molecules. In this embodiment, a reaction-diffusion equation will be used:

[0089]

[0090] in:

[0091] R(c) is a function that takes into account all local responses involving cytokines or signaling molecules, for example, the Fisher equation (Fisher, The Wave of Advantageous Genes, Annals of Eugenics, 1937).

[0092] In medical imaging scans, such as coronary artery CCTA, a continuous media approach can be used to simulate the diffusion of cytokines from the plaque / vessel wall interface into the perivascular adipose tissue.

[0093] The simulation can be initialized at time t=0 by assigning a high concentration value (c) to those voxels corresponding to the plaque and the adjacent vessel wall, and assigning a zero concentration (c) to those voxels corresponding to the perivascular tissue.

[0094] At each time step of the simulation, the concentration of cytokines or signaling molecules in each voxel corresponding to perivascular tissue is updated based on the reaction of molecules within the voxel (if reacting components are included) and the diffusion between adjacent voxels based on concentration differences according to the diffusion equation described above.

[0095] This simulation can be performed a fixed number of iterations, or until the concentration of voxels corresponding to perivascular tissue decreases below a predetermined threshold between consecutive time steps (e.g., reaching a steady state). At this point, the specified concentration threshold can be used to identify which voxels corresponding to perivascular tissue should be included in the region used to quantify radioactivity (i.e., all voxels with concentrations above the threshold).

[0096] Vascular area

[0097] A “vascular region” can be a region of the coronary arteries, carotid arteries, aorta, or any other artery in the body. For example, a vascular region could include the mid-segment of the right coronary artery.

[0098] In a preferred embodiment, the vascular region is limited to the perivascular space surrounding a coronary artery, carotid artery, aorta, or any other artery in the human body. Alternatively, the vascular region may encompass an entire radial segment, including the vessel lumen, vessel wall, and perivascular space.

[0099] The method may further include segmenting the vascular region along the length of the analyzed vessel to facilitate analysis of smaller portions of the vessel in the longitudinal direction. For the avoidance of ambiguity, this longitudinal segmentation differs from radial segmentation of the vascular region—that is, it occurs in a different plane.

[0100] Determine the value of radioactivity density

[0101] Radiometric density, measured in Henle units (HU), is a measure of the relative extent to which X-rays cannot penetrate a material. The term "radiometric density" is synonymous with the term "attenuation," and the two terms are used interchangeably.

[0102] Measuring attenuation values ​​allows for the differentiation of tissue types in CT scans based on their varying radiopaqueness. Fat has a relatively low radioactivity density, typically much lower than that of muscle, blood, and bone. The exact range of attenuation values ​​corresponding to different tissue types usually depends on factors such as CT scan parameters and the type of software used to analyze the medical imaging data.

[0103] For example, the following software program specifies the types of blood vessels and surrounding tissues as follows.

[0104] Coronary plaque analysis 2.0.3 syngo.via FRONTIER, Siemens

[0105] Calcified plaques: >700 HU

[0106] Lumen: 150 – 700 HU

[0107] Non-calcified fibrous plaques: 30 – 150 HU

[0108] Non-calcified lipid-rich: ≤ 30 HU

[0109] QAngioCT version 3.1.3.13 Medis medical imaging system

[0110] Dense calcium: >351 HU

[0111] Fibrotic plaques: 151-350 HU

[0112] Fibrous fatty plaques: 31-150 HU

[0113] Necrotic core: -30-30 HU

[0114] SUREPlaque, version 6.3.2; Vital Images

[0115] Calcified plaques: >150 HU

[0116] Fibrotic plaques: 50 – 150 HU

[0117] Fatty plaque: -100 – 49 HU

[0118] Therefore, technicians in the fields of cardiology and medical imaging are able to distinguish different tissue types from CT image data based on the equipment and software used, without having to provide precise definitions in the HU.

[0119] The term “voxel” has its conventional meaning in this field and is a contraction of the words “volume” and “element”, referring to each of the arrays of discrete volume elements that constitute a conceptual three-dimensional space.

[0120] Determining the value of radial segment radioactivity can include determining the RSA—the average radioactivity or attenuation of all voxels within the radial segment. This can be calculated as the cumulative sum of all voxels within the radial segment divided by the number of voxels in the radial segment.

[0121] In another embodiment, determining the value of segmental radioactivity density may include determining RSA. PV —The average radiometric density or attenuation of all voxels corresponding to the perivascular tissue within the radial segment.

[0122] As used herein, the term "perivascular" refers to the space surrounding a blood vessel. The terms "perivascular tissue" or "perivascular space" refer to the tissue surrounding a blood vessel and may include perivascular adipose tissue (PVAT).

[0123] In one embodiment, the value of radiodensity may involve determining the radiodensity value of one or more specific types of tissue, such as one or more of the following: adipose tissue, fibrous plaque, calcified plaque, or water.

[0124] Preferably, the radiodense value encompasses perivascular tissue in the radial segmental perivascular space. Preferably, the perivascular tissue includes radial segmental perivascular adipose tissue (PVAT). In some embodiments, voxels corresponding to plaques are not included in the radiodense value. Since calcified plaques are generally considered more stable and less prone to rupture than fibrous plaques, voxels corresponding to calcified plaques are preferably not included in the radiodense value.

[0125] In the context of this invention, the "average" value is understood as a central or typical value, which can be calculated from a sample of measurements using formulas widely known and accepted in the art. Preferably, the average value is calculated as the arithmetic mean of a sample of attenuation values, but it can also be calculated as the geometric mean, harmonic mean, median, or mode of a set of collected attenuation values. The average value can be calculated by referring to data collected from all voxels within a concentric tissue layer, or by referring to a selected group of voxels within a concentric tissue layer (e.g., voxels from hydrated, adipose, or fibrous tissue).

[0126] In a preferred embodiment, step (ii) includes determining the radiometric density of a voxel corresponding to perivascular adipose tissue (PVAT) in the radial segmental perivascular region. Based on this measurement, the indices fat attenuation index (FAI) and / or perivascular FAI (FAI) can be determined. PVAT These indicators reflect the standardized weighted average attenuation of the perivascular region surrounding the coronary arteries. They have been found to be sensitive and dynamic biomarkers of coronary inflammation and have been identified as strong and independent predictors of adverse cardiac events.

[0127] FAI and FAI PVAT Based on the understanding that in the presence of vascular inflammation, the release of pro-inflammatory molecules from the diseased vessel wall inhibits the differentiation and lipid accumulation of preadipocytes in the perivascular tissue (PVT), resulting in smaller, less differentiated, and lipid-free adipocytes. This is related to the shift in the radioactivity density (measured by attenuation) of the PVT in computed tomography (CT) imaging from a more negative (close to -190) to a smaller negative (close to -30) Henle units (HU) value, which can be attributed to FAI and FAI. PVAT The biological significance and clinical value of FAI have been extensively validated. These biomarkers are described in detail in WO 2016 / 024128A1 and WO 2018 / 078395 A1, the contents of which are incorporated herein by reference.

[0128] will FAI PVAT The FAI score is limited to radial segments of the vessel, rather than averaging over the entire vessel or the space around the vessel, which makes it possible to detect small but highly inflammatory “hot spots” that would otherwise be missed due to averaging over a larger area.

[0129] Medical imaging data

[0130] Medical imaging data can be radiographic data. In a preferred embodiment, medical imaging data can be computed tomography (CT) data.

[0131] CT scans are preferably performed along the length of the blood vessel, for example, along the length of the right coronary artery, left anterior descending artery, left circumflex artery, aorta, carotid artery, or femoral artery.

[0132] To avoid ambiguity, the method of the present invention utilizes CT scan data obtained by scanning a living organism, but the claimed method is not performed on a living human or animal body. Therefore, the method of the present invention is non-invasive and based on the analysis of conventional CT images; it requires no additional image acquisition.

[0133] Vascular inflammation

[0134] The output value determined in step (b) can provide a measure of inflammation in the vascular region. Vascular inflammation refers to a progressive inflammatory condition characterized by leukocyte infiltration of blood vessels and the accumulation of atherosclerotic plaques within the vessel walls (particularly within arterial walls).

[0135] Vascular inflammation is a key feature of atherosclerosis formation and plays a crucial role in atherosclerotic plaque stability by triggering plaque rupture that leads to acute coronary syndrome (see Ross R. N EN). g (See J Med 1999;340:115-26, and Major AS et al. Circulation 2011;124:2809-11). Importantly, more than 50% of acute coronary syndromes are caused by highly inflammatory but anatomically insignificant atherosclerotic plaques (Fishbein MC et al. Circulation 1996;94:2662-6), which may be missed by less sensitive methods.

[0136] The method of this invention enables the detection of subclinical plaques, preplaque states, and / or early atherosclerotic disease activity in the vessel walls of inflamed vascular regions—thanks to improved sensitivity and more targeted assessment of the pathology or condition of vascular regions. Furthermore, the method can identify subclinical extensions of plaques already visible on medical imaging data. Therefore, the method can identify tissue that is transforming into high-risk fatty plaque tissue but extends beyond the currently visible clinical margin, thereby enabling earlier risk detection and intervention.

[0137] The method according to the first aspect may also include the step of identifying inflammatory plaques within the vascular region. For example, if plaques are present in a radial segment, such as... Figure 3 For plaque 305, the output determined in step (b) can be used to characterize the plaque as inflammatory. For high-risk fatty plaques (which may be non-calcified), the entire plaque can be classified as inflammatory using this technique.

[0138] On the other hand, calcified plaques tend to be more stable and have a lower risk of rupture, which can lead to adverse cardiac events. However, they may contain inflammatory, non-calcified fatty areas that may be prone to rupture.

[0139] Therefore, the method of the present invention may include the step of identifying the inflammatory portion of a plaque within a vascular region. For example, refer to Figure 3 In vessel wall 302, plaque 205 spans four independent radial segments of a predefined geometry. If the output value determined in step (b) indicates that a particular segment containing the plaque is particularly inflammatory, then the portion of the plaque contained in that segment can be classified as inflammatory. Therefore, this method allows for the mapping of unstable or inflammatory heat maps on the plaque.

[0140] On the other hand, there may be segments containing plaque that do not show signs of inflammation, making the plaque areas within these segments potentially classifiable as low-risk. This is expected, for example, for calcified plaques that are unlikely to cause adverse cardiac events.

[0141] Therefore, the method of the present invention can classify plaques or inflammatory plaques into low-risk, medium-risk, high-risk, or very high-risk categories.

[0142] plaque characterization

[0143] According to a second aspect, the present invention provides a method for characterizing plaques in the blood vessels of a subject, the method comprising:

[0144] (a) Providing medical imaging data acquired along the length of the blood vessel; and

[0145] (i) Identify voxels corresponding to plaques by their radioactivity;

[0146] (ii) Determine radial segments of the blood vessel containing voxels corresponding to the plaque, wherein the radial segments extend from a point in the center of the blood vessel to a distance extending beyond the outer wall of the blood vessel;

[0147] (iii) Determine the values ​​of the radioactivity density and / or one or more radiomics characteristics of the radial segment;

[0148] (b) Compare each value determined in (iii) with a predetermined threshold, or use the absolute value determined in (iii) to generate an output value indicating inflammation of the plaque.

[0149] At least one method step is implemented by a computer.

[0150] According to a second aspect of the invention, step (ii) may include determining the radiometric density of a voxel corresponding to perivascular adipose tissue (PVAT) in a radial segment, the utility of which has been described in detail above.

[0151] Radial segments can be:

[0152] a) Predefined geometry; or

[0153] b) The radius of the radial segment is defined by the physical limit of the voxel corresponding to the patch.

[0154] The predefined geometry can be a portion of the cross-section of the blood vessel in the radial direction. For example, the angle of the radial segment from a point in the center of the blood vessel can be 1° to 359°, preferably 2° to 270°, more preferably 10° to 180°, and even more preferably 30° to 150°.

[0155] Alternatively, the radius of the radial segment can be defined by the physical limit of the voxel corresponding to the plaque, such that the plaque contained within the vessel wall is completely contained within a given cross-section or along the length of the vessel by the radial segment. For example, if the plaque is located on a quarter of the vessel, the radial segment could be a quarter of the vessel and would contain the voxel corresponding to the plaque. Figure 1B As shown.

[0156] The voxels corresponding to the plaques may include calcified and non-calcified plaques. Preferably, the plaques are non-calcified, as non-calcified plaques are of higher risk.

[0157] In some embodiments, when determining the values ​​of radiodensity and / or one or more radiomics features in step (iii), voxels corresponding to calcified plaques are not included in the analysis and are not used to determine the output values ​​in step (b).

[0158] Inflammatory plaques can be fibrous or non-fibrous; and they can be further characterized or classified as one or more of the following: hotspot, unstable, high-risk, and potential culprit lesions.

[0159] When a plaque is characterized as inflammatory in step (b), the subject may be given a therapy to treat the plaque or prevent its progression.

[0160] In step (iii), the values ​​of radiodensity and / or one or more radiomics features may be taken from the entire radial segment, including the lumen, vessel wall and perivascular space.

[0161] Preferably, the value is taken from the vessel wall and the perivascular space. Alternatively, the values ​​of radioactivity and / or one or more radiomics characteristics are taken from the perivascular space.

[0162] The preferred scheme for the radiation density characteristics is the same as in the first aspect.

[0163] Radiomics characteristics

[0164] Step (ii) of the first aspect and step (iii) of the second aspect of the invention may include determining values ​​for at least two radiomics features from medical imaging data. The values ​​for each radiomics feature are then included in step (b) to generate output values ​​indicating pathology or condition or plaque inflammation in a vascular region.

[0165] In WO 2020 / 058713 A1, it was determined that the use of radiomics features adds incremental value to traditional risk factors and established CCTA risk classification tools in predicting future adverse cardiovascular events and assessing cardiovascular health and risk, and further helps to detect the presence of vascular inflammation, local plaque inflammation, and unstable coronary artery lesions.

[0166] Preferably, in addition to one or more, preferably two or more, radiomics features, the radiodensity of the vascular segment is also determined. Using two or more radiomics features provides a “radiomics signature spectrum” and offers a tool for further characterizing vascular segments or plaques within radial segments.

[0167] In a preferred embodiment, radiomics features are taken from the perivascular space within a radial segment.

[0168] Particularly preferably, in addition to radiomics characteristics, the radiodensity value of the vascular region (which may be limited to the space around the blood vessel) is also determined.

[0169] For an embodiment that determines one or more radiomics features, the output value determined in step (b) can provide a measure of the texture of the vascular region.

[0170] If texture indication is to be provided, at least one radiomics feature can provide a measure of texture. If more than one radiomics feature has been identified, each radiomics feature can provide a measure of radial segment texture (i.e., each of at least two radiomics features can be a texture statistic). Alternatively, one or more radiomics features may be required to provide a measure of texture.

[0171] In a preferred embodiment, the method of the present invention includes determining two or more radiomics features. More preferably, three or more, or four or more radiomics features. Radiomics features may be determined based on the perivascular region of a vascular segment. Alternatively, radiomics features may be determined based on the entire radial segment (including the vessel containing the lumen and vessel wall).

[0172] If two or more radiomics features of a radial segment are determined, the radiomics feature spectrum of the radial segment can be calculated. The radiomics feature spectrum is described in detail in WO 2020 / 058713, the entire contents of which are incorporated herein by reference.

[0173] In deriving a radiomics signature spectrum, the method may include using datasets, particularly radiomics datasets, to construct a radiomics signature spectrum or score. This score can then be used to characterize the pathology or condition of vascular regions, characterize plaque inflammation, provide cardiovascular risk scores, or guide drug treatment.

[0174] A radiomics feature spectrum can be calculated based on a (second) plurality of perivascular radiomics features. The dataset may contain measurements of a (first) plurality of perivascular radiomics features of the perivascular region obtained from medical imaging data of each of a plurality of individuals. The plurality of individuals may include a first group of individuals who have met clinical endpoints indicative of cardiovascular risk and / or specific biological states (such as inflammation, fibrosis, edema, vascular distribution, or combinations thereof); and a second group of individuals who have not yet met clinical endpoints indicative of cardiovascular risk and / or specific biological states. The second plurality of radiomics features may be selected from the first plurality of radiomics features, particularly providing a radiomics feature spectrum for predicting cardiovascular risk, as determined from the dataset or using the dataset (e.g., using machine learning algorithms). Therefore, the radiomics feature spectrum can be calculated based on further radiomics features (e.g., selected from the (first) plurality of radiomics features) in addition to at least two radiomics features.

[0175] Radiomic features can be selected from radiomic features of clusters 1 to 9, where:

[0176] Cluster 1 consists of short-run high gray level emphasis, high gray level emphasis, high gray level run emphasis, autocorrelation, average, joint average and high gray level region emphasis;

[0177] Cluster 2 consists of skewness, skewness LLL, kurtosis, 90th percentile, 90th percentile LLL, median LLL, kurtosis LLL, and median;

[0178] Cluster 3 consists of run entropy, dependency entropy LLL, dependency entropy, region entropy LLL, run entropy LLL, and average value LLL;

[0179] Cluster 4 consists of small-area low-gray-level emphasis, low-gray-level area emphasis, short-run low-gray-level emphasis, low-gray-level run emphasis, low-gray-level emphasis, low-dependency low-gray-level emphasis, gray-level variance LLL (GLSZM), gray-level variance (GLDM), variance, gray-level variance (GLDM), difference variance LLL, gray-level variance LLL (GLRLM), variance LLL, gray-level variance LLL (GLDM), sum of squares, contrast LLL, mean absolute deviation, interquartile range, robust mean absolute deviation, long-run low-gray-level emphasis, difference variance, gray-level variance (GLSZM), normalized inverse difference moment, mean absolute deviation LLL, sum of squares LLL, and contrast.

[0180] Cluster 5 consists of regional entropy, normalized gray level nonuniformity (GLRLM), normalized gray level nonuniformity LLL (GLRLM), and entropy, joint energy, entropy, normalized gray level nonuniformity (GLDM), joint energy, normalized gray level nonuniformity LLL (GLDM), uniformity LLL, and entropy LLL and uniformity.

[0181] Cluster 6 consists of regional entropy HHH, normalized size regional non-uniformity HHH, and small region emphasis HHH;

[0182] Cluster 7 consists of strength, roughness HLL, roughness, roughness LHL, roughness LLL, roughness LLH, roughness HHH, roughness HLH, roughness HHL and roughness LHH;

[0183] Cluster 8 consists of cluster tendency (LLL), cluster tendency, sum of squares (LLL), mean absolute deviation (LLL), gray level variance (GLDM), variance (LLL), gray level variance (GLRLM), gray level variance (GLRLM), robust mean absolute deviation (LLL), gray level variance (GLDM), variance, mean absolute deviation, cluster salientity, sum of entropy (LLL), interquartile range (LLL), gray level variance (GLSZM), sum of squares, robust mean absolute deviation, sum of entropy, interquartile range, cluster salientity (LLL), entropy (LLL), 10th percentile (LLL), 10th percentile; and

[0184] Cluster 9 consists of size region nonuniformity LLL, dependent nonuniformity HLL, gray level nonuniformity HLL (GLSZM), gray level nonuniformity (GLSZM), run length nonuniformity HHL, run length nonuniformity LHL, dependent nonuniformity LHL, dependent nonuniformity, run length nonuniformity HLH, busyness, run length nonuniformity LLH, dependent nonuniformity LLH, dependent nonuniformity LLL, size region nonuniformity, energy HLL, run length nonuniformity LHH, size region nonuniformity HLL, gray level nonuniformity LLH (GLSZM), gray level nonuniformity LHL (GLSZM), gray level nonuniformity LLL (GLSZM), run length nonuniformity HLL, gray level nonuniformity HLH (GLSZM), gray level nonuniformity HHL (GLSZM), run length nonuniformity and run length nonuniformity HHH.

[0185] Preferably, at least two radiomics features are selected; preferably, the at least two radiomics features are each selected from different clusters.

[0186] The definitions of radiomics features used in this article are generally well understood in the field by their names. However, for ease of reference, the definitions of the features used in this article are provided in Tables R1 to R7 below. The radiomics features in Tables R1 to R7 are defined according to the radiomics features used in the Pyradiomics package (http: / / pyradiomics.readthedocs.io / en / latest / features.html, see van Griethuysen, JJM, Fedorov, A., Parmar, C., Hosny, A., Aucoin, N., Narayan, V., Beets-Tan, RGH, Fillon-Robin, JC, Pieper, S., Aerts, HJWL (2017). Computational Radiomics System to Decode the Radiographic Phenotype. CancerResearch, 77(21), e104–e107. https: / / doi.org / 10.1158 / 0008-5472.CAN-17-0339). Most features defined in Tables R1 through R7 conform to the feature definitions described in the Imaging Biomarker Standardization Initiative (IBSI), which is available in Zwanenburg et al. (2016). (Zwanenburg, A., Leger, S., Vallières, M., and Löck, S. (2016). Image biomarker standardization initiative - feature definitions. In eprint arXiv:1612.07003 [cs.CV]). If the definitions provided below are not entirely consistent with the IBSI definitions, it should be understood that the definitions of either aspect can be used according to the present invention. Ultimately, the precise mathematical definition of radiomics features is not critical, as minor modifications will not affect the general properties of the image measured by each feature. Therefore, minor modifications to features (e.g., adding or subtracting constants or scaling) and alternative definitions of features are intended to be included in the present invention.

[0187] a. First-order statistics

[0188] These statistics describe the central tendency, variability, homogeneity, asymmetry, skewness, and amplitude of attenuation values ​​within a given region of interest (ROI), regardless of the spatial relationships of individual voxels. Therefore, they describe the quantitative and qualitative characteristics of the entire ROI (PVR). A total of 19 features were calculated for each of the eight wavelet transforms and the original CT image, as follows:

[0189] set up:

[0190] X represents the N contained in the ROI. p Attenuation or radiometric density value of a voxel (e.g., in HU).

[0191] P(i) is a subset of N g A first-order histogram of discrete intensity levels, where N g This represents the number of non-zero bins, starting from 0 with equal spacing and a width of a certain value.

[0192] p(i) is a normalized first-order histogram, equal to

[0193] c is an offset intensity value to prevent negative values ​​in X. This ensures that the voxel with the lowest gray value contributes the least to the energy, rather than the voxel with the gray level intensity closest to 0. Since the HU range (-190 to -30 HU) of adipose tissue (AT) in PVR does not include zero, c can be set to c=0. Therefore, higher energy corresponds to lower radiometric density AT, and thus higher lipophilicity.

[0194] It is an arbitrarily small positive number (e.g., ≈ 2.2 × 10⁻⁶). -16 )

[0195]

[0196]

[0197]

[0198] b. Shape-related statistics

[0199] Shape-related statistics describe the size and shape of a given region of interest (ROI) without considering its voxel attenuation. Since they are independent of gray-level intensity, shape-related statistics are consistent across all wavelet transforms and the original CT images, and therefore are calculated only once. They are defined as follows:

[0200] set up:

[0201] V represents the volume of the ROI, in mm. 3

[0202] A represents the surface area of ​​the ROI, in mm. 2

[0203]

[0204]

[0205] c. Gray-level co-occurrence matrix (GLCM)

[0206] In short, GLCM describes the number of times a voxel with a given decay value i is adjacent to a voxel with a value j. The size is N. g ×N g The GLCM describes the second-order joint probability function of an image region constrained by a mask, and is defined as P(i,j|δ,θ). The (i,j)th element of this matrix represents the number of times a combination of gray levels i and j occurs in two pixels in the image, which are δ pixels apart along an angle θ. The distance δ from the central voxel is defined according to the infinity norm. For δ=1, this results in 2 neighbors for each of the 13 angles in 3D (26-connectivity), and for δ=2, it results in 98-connectivity (49 unique angles). To obtain rotation-invariant results, the statistics are calculated in all directions and then averaged to ensure a symmetric GLCM.

[0207] set up:

[0208] It is an arbitrarily small positive number (e.g., ≈2.2 × 10⁻� -16 )

[0209] P(i,j) is the co-occurrence matrix of any δ and θ.

[0210] p(i,j) is the normalized co-occurrence matrix, equal to

[0211] N g It is the number of discrete intensity levels in the image.

[0212] Marginal row probability

[0213] Marginal column probability

[0214] μ x For p x The average gray level intensity is defined as

[0215] μ y For p y The average gray level intensity is defined as

[0216] σ xFor p x Standard deviation

[0217] σ y For p y Standard deviation

[0218] Where "i+j=k," and "k=2,3,…,2N_g

[0219] Where "|ij|=k," and "k=0,1,…,N g -1

[0220] For p x entropy

[0221] For p y entropy

[0222]

[0223] For distance-weighted calculations, the GLCM matrix is ​​weighted by a weighting factor W, then summed and normalized. The weighting factor W is calculated based on the distance between adjacent voxels. The result is calculated, where d is the distance at the relevant angle.

[0224]

[0225]

[0226]

[0227]

[0228] d. Gray-scale size region matrix (GLSZM)

[0229] Gray-scale size region (GLSZM) describes gray-scale regions within a Region of Interest (ROI) and is defined as the number of connected voxels sharing the same gray-scale intensity. Voxels are considered connected if the distance is based on an infinity norm of 1 (26 connected regions in 3D and 8 connected regions in 2D). In the gray-scale size region matrix P(i, j), the (i, j) elements equal the number of regions in the image where gray-scale i and size j occur. Unlike GLCM and GLRLM, GLSZM is rotation-independent, calculating only one matrix in all directions within the ROI.

[0230] set up:

[0231] N g The number of discrete intensity values ​​in the image

[0232] N s The number of discrete regions in the image

[0233] N p The number of voxels in the image

[0234] N z The number of regions in the ROI is equal to And 1 ≤ N z ≤ N p

[0235] P(i, j) is the size region matrix.

[0236] p(i, j) is the normalized size region matrix, defined as follows:

[0237] It is an arbitrarily small positive number (e.g., ≈2.2 × 10⁻� -16 ).

[0238]

[0239]

[0240]

[0241] e. Gray-level run-length matrix (GLRLM)

[0242] The Gray Level Run-Length Matrix (GLRLM) describes gray level runs, which are defined as the length (in pixels) of consecutive pixels with the same gray level value. In the gray level run-length matrix P(i, j|θ), the (i, j)th element describes the number of runs of gray level i and length j along angle θ in the image (ROI).

[0243] set up:

[0244] N g The number of discrete intensity values ​​in the image

[0245] N r The number of discrete run lengths in the image

[0246] N p The number of voxels in the image

[0247] N z (θ) represents the number of runs in the graph along angle θ, equal to And 1≤N z (θ)≤N p

[0248] P(i,j|θ) is the run length matrix in any direction θ.

[0249] p(i,j|θ) is the normalized run-length matrix, defined as follows:

[0250] It is an arbitrarily small positive number (e.g., ≈2.2 × 10⁻� -16 ).

[0251] By default, eigenvalues ​​are calculated separately for each angle using GLRLM, and the average of these values ​​is returned. If distance weighting is enabled, GLRLM is weighted by the distances between adjacent voxels, then summed and normalized. The eigenvalues ​​are then calculated on the resulting matrix. The distances between adjacent voxels are calculated for each angle using the norm specified in "weightingNorm".

[0252]

[0253]

[0254] f. Neighborhood Gray-Level Difference Matrix (NGTDM) Features

[0255] The neighborhood gray-level difference matrix quantifies the difference between a gray-level value and the average gray-level value of its neighbors within a distance δ. The sum of the absolute differences of gray-level i is stored in the matrix. Let X... gl To segment the voxel set, For position If the gray level of a voxel is given, then the average gray level of its neighborhood is:

[0256]

[0257] Here, W is a neighborhood that also belongs to X. gl The number of voxels.

[0258] set up:

[0259] n i For X gl Number of voxels with gray level i

[0260] Nv,p is X gl The total number of voxels is equal to ∑n i (i.e., the number of voxels with a valid region; having at least one neighbor). , where N p It represents the total number of voxels in the ROI.

[0261] p i The probability of gray level is equal to

[0262] The sum of the absolute differences of gray levels i

[0263] N g Discrete gray levels

[0264] N g p is p i Gray levels ≠ 0

[0265]

[0266] g. Gray-level dependency matrix (GLDM)

[0267] The Gray-Level Dependency Matrix (GLDM) quantifies gray-level dependencies in an image. Gray-level dependency is defined as the number of connected voxels within a distance δ that depend on a central voxel. If |i−j|≤α, then a neighboring voxel of gray level j is considered dependent on a central voxel of gray level i. In the gray-level dependency matrix P(i,j), the (i,j)th element describes the number of times a voxel of gray level i appears in the image and has j dependent voxels in its neighborhood.

[0268] set up:

[0269] N g The number of discrete intensity values ​​in the image

[0270] N d The number of discrete dependencies in the image

[0271] N z The number of dependent regions in the image is equal to

[0272] P(i,j) is the dependency matrix

[0273] p(i,j) is the normalized dependency matrix, defined as follows:

[0274]

[0275]

[0276] Output value

[0277] To determine output values ​​that can be used to characterize the pathology or condition of a vascular region or to characterize plaque inflammation (in step (b) of the first and second aspects of the invention), the values ​​of the radial segment radioactivity and / or one or more radiomics features are compared with predetermined thresholds. Alternatively, absolute values ​​are used.

[0278] The threshold can be a distribution of values, and more preferably, the threshold is vascular-specific and / or specific to how the value is derived along the proximal or distal end of the blood vessel.

[0279] Particularly preferably, the threshold and / or output value are adjusted for biological factors that may affect the expected value. Preferably, the output value is adjusted. The biological factors are preferably selected from one or more of the following: the subject's age, the subject's sex, the subject's ethnicity, background adipocyte size, partial volume effect, and vascular type.

[0280] The partial volume effect can be determined using the expectation-maximization method to iteratively refine the estimated “partial volume correction” voxel configuration, which most closely matches the observed voxel configuration when convolved with the point spread function of the imaging system.

[0281] Technical factors can also affect the obtained expected radiation density and / or radiomics values. Therefore, predetermined thresholds and / or output values ​​can be adjusted for these technical factors, such as one or more of the following: CT scanner tube voltage, reconstruction algorithm, iodinated contrast agent, contrast agent type, injection rate, aortic enhancement, left ventricular blood pool enhancement, signal-to-noise ratio, contrast-to-noise ratio, mA, cardiac gating method, single-energy and multi-energy image acquisition, CT scanner type, heart rate, heart rhythm, or blood pressure.

[0282] Corrections for biological and / or technical factors are preferably performed by a processing system (i.e., by computer). Such a processing system can identify the need for these corrections and perform them automatically.

[0283] The coefficients of each variable used to generate the output values ​​in step (b) can be derived from a Cox hazard regression model. Alternatively, the cutoff point can be derived from a receiver operating characteristic (ROC) curve. Correction can also be determined using linear or nonlinear calibration factors derived from repeated scans of the same patient (or phantom subject) under different scanner settings. Alternatively, a neural network-based approach can be used.

[0284] Computer implementation

[0285] At least one step of the method according to the first and second aspects of the present invention is implemented by a computer. Therefore, the present invention requires a processing system. Particularly preferably, at least two steps are implemented by a computer. Preferably, the entire method is implemented by a computer, thereby providing a computer-implemented method according to the first and second aspects for characterizing the pathology or condition of a vascular region of a subject, and a computer-implemented method for characterizing plaque in the blood vessels of a subject.

[0286] Most preferably, each step is performed by the processing system, so that the method of the present invention is performed automatically given input medical imaging data without manual intervention. However, for quality control purposes, it is envisioned that a trained operator can check and adjust specific parameters (e.g., a point in the center of the blood vessel, the centerline of the blood vessel, or a predetermined distance from the outer wall of the blood vessel) to ensure compliance with established clinical practice.

[0287] Therefore, the method of the present invention can be implemented automatically using dedicated software, providing a rapid, non-invasive assessment of individual vascular pathology / conditions and risk profiles and the likelihood of adverse events, and guiding clinical decision-making.

[0288] In a third aspect, the present invention provides a computer program product including executable instructions that, when executed by one or more processors, cause the one or more processors to perform the method of the first aspect or the second aspect.

[0289] In a fourth aspect, a system is provided, comprising one or more processors configured to perform the methods of the first or second aspect.

[0290] A non-transitory computer-readable medium is further disclosed, comprising executable instructions that, when executed by one or more processors, cause the one or more processors to perform the method according to the first aspect or the second aspect.

[0291] Figure 4 A system example suitable for implementing the embodiments of the method is schematically illustrated. System 1130 includes at least one server 1110 that communicates with a reference data repository 1120. The server may also communicate with other hardware operated by healthcare professionals, such as via a communication network 1140.

[0292] In some embodiments, the server may obtain (e.g., using a reference data repository) predetermined thresholds that can be corrected for biological and technical factors. The server may then provide corrected output values ​​according to the methods described herein, characterizing the pathology or condition of vascular segments and / or plaque inflammation, to provide a useful clinical picture of the subject's current vascular health and risk of adverse cardiac events.

[0293] A suitable example of server 1110 is as follows Figure 5 As shown. In this example, the server includes at least one microprocessor 1200, memory 1201, optional input / output devices 1202 (such as a keyboard and / or display), and an external interface 1203, interconnected via bus 1204 as shown. In this example, the external interface 1203 can be used to connect the server 1110 to peripheral devices, such as a communication network 1140, a reference data repository 1120, other storage devices, etc.

[0294] Although a single external interface 1203 is shown, this is for illustrative purposes only; in reality, multiple interfaces can be provided using various methods such as Ethernet, serial, USB, wireless, etc.

[0295] In use, the microprocessor 1200 executes instructions in the form of application software stored in memory 1201 to allow the execution of desired processes, including communicating with the reference data repository 1120 to receive and process input data, and providing a preferred corrected output score according to the methods described above. The application software may include one or more software modules and may execute in a suitable execution environment, such as an operating system environment.

[0296] Therefore, it should be understood that server 1200 can be formed from any suitable processing system, such as a properly programmed client device, PC, web server, network server, etc. In a particular example, server 1200 is a standard processing system, such as an Intel architecture-based processing system, which executes software applications stored on non-volatile (e.g., hard disk) storage, although this is not required. However, it should also be understood that the processing system can be any electronic processing device, such as a microprocessor, microchip processor, logic gate configuration, firmware (optionally associated with the implementation logic, such as an FPGA (Field Programmable Gate Array)), or any other electronic device, system, or arrangement. Therefore, although the term server is used, this is for illustrative purposes only and is not intended to be limiting.

[0297] Although server 1200 is shown as a single entity, it should be understood that server 1200 may be distributed across multiple geographically separated locations, for example, by using a processing system and / or database 1201 provided as part of a cloud-based environment. Therefore, the above arrangement is not required, and other suitable configurations can be used. In a preferred embodiment, the processing system is cloud-based.

[0298] Risk Score

[0299] The invention envisions established biomarkers of vascular inflammation and biological states (such as fibrosis, edema, and vascular distribution), as well as established risk factors, which can be combined with the output values ​​of the invention to provide a more complete risk score for a given subject.

[0300] In a preferred embodiment, step (b) of the method of the present invention includes determining one or more of the following vascular features:

[0301] (i) Calcium index (Calcium-i);

[0302] (ii) Perivascular water index (PVWi);

[0303] (iii) Epicardial adipose tissue volume (EpAT-vol);

[0304] (iv) Fat decay index of epicardial adipose tissue (FAI) EpAT );

[0305] (v) Fat attenuation index (FAI) of perivascular adipose tissue PVAT );

[0306] (vi) Fiber plaque index (FPi);

[0307] (vii) Perivascular water index (PVWi);

[0308] (viii) Perivascular volume characterization index (VPCI);

[0309] (ix) Presence, volume, or radiomics profile of plaques;

[0310] (x) The presence, volume, or radiomics profile of high-risk plaques;

[0311] (xi) Presence, volume, or radiomics profile of low-attenuation patches;

[0312] (xii) Presence, volume, or radiomics profile of lipid-rich plaques;

[0313] (xiii) Presence, volume, or radiomics profile of fibrous plaques;

[0314] (xiv) Presence, volume, or radiomics profile of non-calcified plaques;

[0315] (xv) Presence, volume, or radiomics profile of calcified plaques; and

[0316] (xvi) Blood vessel volume, diameter, cross-sectional area, surface area, length, location, or remodeling.

[0317] One or more vascular features can be identified for one or more blood vessels. The calculated risk may include: determining the number of blood vessels that have one or more vascular features present, or whose numbers are above or below a specified threshold.

[0318] Step (b) may also include determining one or more of the following plaque characteristics:

[0319] (xvii) Patch density;

[0320] (xviii) constitutes;

[0321] (xix) Calcification;

[0322] (xx) radioactivity density;

[0323] (xxi) position;

[0324] (xxii) Volume;

[0325] (xxiii) Surface area;

[0326] (xxiv) Geometric shape;

[0327] (xxv) heterogeneity;

[0328] (xxvi) diffusivity; and

[0329] (xxvii) The ratio of volume to surface area.

[0330] Preferably, the values ​​(i)-(xxvii) above are taken from the radial segments and used in step (b) to generate output values ​​characterizing radial segment pathology or status or characterizing inflammation of any plaque (if present).

[0331] Step (b) of the method may also include considering one or more of the following risk factors or subject characteristics:

[0332] (xxviii) Age of the subject;

[0333] (xxix) The gender of the subject;

[0334] (xxx) The race / ethnicity of the subject;

[0335] (xxxi) Coronary artery calcium;

[0336] (xxxii) Hypertension;

[0337] (xxxiii) Hyperlipidemia / Hypercholesterolemia;

[0338] (xxxiv) Diabetes mellitus;

[0339] (xxxv) The presence of coronary artery disease;

[0340] (xxxvi) To smoke;

[0341] (xxxvii) Family history of heart disease; and

[0342] (xxxviii) Hereditary status.

[0343] The definitions of these biomarkers are known to those skilled in the art and are extensively described in WO 2018 / 078395 A1, the contents of which are incorporated herein by reference.

[0344] According to a first aspect of the invention, a method for characterizing the pathology or condition of a vascular region of a subject is further provided according to aspects 1-1 to 1-18.

[0345] Aspect 1-1: A method for characterizing the pathology or condition of a vascular region in a subject, wherein:

[0346] (i) Determine the value of the radioactivity density in step (ii);

[0347] (ii) The vascular region includes the space surrounding the blood vessel;

[0348] (iii) The method is implemented by a computer.

[0349] Aspect 1-2: A method for characterizing the pathology or condition of a vascular region in a subject, wherein:

[0350] (i) In step (ii), determine the values ​​of radiodensity and one or more radiomics features;

[0351] (ii) The vascular region includes the space surrounding the blood vessel;

[0352] (iii) The method is implemented by a computer.

[0353] Aspects 1-3: A method for characterizing the pathology or condition of a vascular region in a subject, wherein:

[0354] (i) Determine the value of the radioactivity density in step (ii);

[0355] (ii) The vascular region comprises the entire radial segment, including the lumen, the vessel wall, and the perivascular space;

[0356] (iii) The method is implemented by a computer.

[0357] Aspects 1-4: A method for characterizing the pathology or condition of a vascular region in a subject, wherein:

[0358] (i) Determine the value of the radioactivity density in step (ii);

[0359] (ii) The vascular region includes the vessel wall and the space surrounding the vessel;

[0360] (iii) The method is implemented by a computer.

[0361] Aspects 1-5: A method for characterizing the pathology or condition of a vascular region in a subject, wherein:

[0362] (i) In step (ii), determine the values ​​of radioactivity density and two or more radiomics characteristics;

[0363] (ii) The vascular region includes the vessel wall and the space surrounding the vessel;

[0364] (iii) The method is implemented by a computer.

[0365] Aspects 1-6: A method for characterizing the pathology or condition of a vascular region in a subject, wherein:

[0366] (i) Determine the value of the radioactivity density in step (ii);

[0367] (ii) The vascular region includes the vessel wall and the space surrounding the vessel;

[0368] (iii) The method is implemented by a computer;

[0369] (iv) The predetermined distance is determined by the processing system using a reference dataset based on the radiometric density in the image or the patient's anatomical structure.

[0370] Aspects 1-7: A method for characterizing the pathology or condition of a vascular region in a subject, wherein:

[0371] (i) In step (ii), determine the values ​​of radioactivity density and two or more radiomics characteristics;

[0372] (ii) The vascular region includes the vessel wall and the space surrounding the vessel;

[0373] (iii) The method is implemented by a computer;

[0374] (iv) The predetermined distance is determined by the processing system using a reference dataset based on the radiometric density in the image or the patient's anatomical structure;

[0375] (v) The radial segment is at an angle of 2 to 270° from a point in the center of the blood vessel.

[0376] Aspects 1-8: A method for characterizing the pathology or condition of a vascular region in a subject, wherein:

[0377] (i) In step (ii), determine the values ​​of radioactivity density and two or more radiomics characteristics;

[0378] (ii) The vascular region includes the vessel wall and the space surrounding the vessel;

[0379] (iii) The method is implemented by a computer;

[0380] (iv) The predetermined distance is determined by the processing system using a reference dataset based on the radiometric density in the image or the patient's anatomical structure;

[0381] (v) The radial segment is at an angle of 2 to 270° from a point in the center of the blood vessel;

[0382] (vi) The value of the radiometric density includes FAI. PVAT .

[0383] According to a second aspect of the invention, a method for characterizing plaques in the blood vessels of a subject is further provided, according to aspects 2-1 to 2-8.

[0384] Aspect 2-1: A method for characterizing plaque in the blood vessels of a subject, wherein:

[0385] (i) Determine the value of the radioactivity density in step (iii);

[0386] (ii) The radiometric density value is taken from the entire radial segment;

[0387] (iii) The method is implemented by a computer.

[0388] Aspect 2-2: A method for characterizing plaque in the blood vessels of a subject, wherein:

[0389] (i) In step (iii), determine the values ​​of radiodensity and one or more radiomics features;

[0390] (ii) The values ​​of the radiodensity and one or more radiomics features are taken from the perivascular space of the radial segment;

[0391] (iii) The method is implemented by a computer.

[0392] Aspect 2-3: A method for characterizing plaque in the blood vessels of a subject, wherein:

[0393] (i) Determine the value of the radioactivity density in step (iii);

[0394] (ii) The value of the radiometric density is taken from the entire radial segment, including the lumen, vessel wall and perivascular space;

[0395] (iii) The method is implemented by a computer.

[0396] Aspects 2-4: A method for characterizing plaque in the blood vessels of a subject, wherein:

[0397] (i) Determine the value of the radioactivity density in step (iii);

[0398] (ii) The value of the radiometric density is taken from the vessel wall and the space surrounding the vessel in the radial segment;

[0399] (iii) The method is implemented by a computer.

[0400] Aspects 2-5: A method for characterizing plaque in the blood vessels of a subject, wherein:

[0401] (i) In step (iii), determine the values ​​of radioactivity density and two or more radiomics characteristics;

[0402] (ii) The values ​​of the radiometric density and two or more radiomics features are taken from the radial segment of the vessel wall and the perivascular space;

[0403] (iii) The method is implemented by a computer.

[0404] Aspects 2-6: A method for characterizing plaque in the blood vessels of a subject, wherein:

[0405] (i) Determine the value of the radioactivity density in step (iii);

[0406] (ii) The value of the radiometric density is determined based on voxels corresponding to the vessel wall and the space surrounding the vessel;

[0407] (iii) The method is implemented by a computer;

[0408] (iv) The predetermined distance is determined by the processing system using a reference dataset based on the radiometric density in the image or the patient's anatomical structure;

[0409] (v) The radial segment is at an angle of 2 to 270° from a point in the center of the blood vessel.

[0410] Aspects 2-7: A method for characterizing plaque in the blood vessels of a subject, wherein:

[0411] (i) In step (iii), determine the values ​​of radioactivity density and two or more radiomics characteristics;

[0412] (ii) The plaque is non-calcified;

[0413] (iii) The method is implemented by a computer;

[0414] (iv) The predetermined distance is determined by the processing system using a reference dataset based on the radiometric density in the image or the patient's anatomical structure;

[0415] (v) The radius of the radial segment is defined by the physical limit of the voxel corresponding to the patch.

[0416] Aspects 2-8: A method for characterizing plaque in the blood vessels of a subject, wherein:

[0417] (i) In step (iii), determine the values ​​of radioactivity density and two or more radiomics characteristics;

[0418] (ii) The plaque is non-calcified;

[0419] (iii) The radius of the radial segment is defined by the physical limit of the voxel corresponding to the patch;

[0420] (iv) The method is implemented by a computer;

[0421] (v) The predetermined distance is determined by the processing system using a reference dataset based on the radiometric density in the image or the patient's anatomical structure;

[0422] (vi) The radial segment is at an angle of 2 to 270° from a point in the center of the blood vessel;

[0423] (vii) The value of the radiometric density is FAI. PVAT .

[0424] Example

[0425] Example 1

[0426] A digital phantom of the coronary artery was constructed using the Python software libraries NumPy, SciPy, and Matplotlib. Figure 6 Focal "inflammatory" areas were added to the phantom. Figure 6 (b) The intensity of the inflammatory region relative to the background PVAT is increased in increments of 2 HU. Random Gaussian noise and image blur are applied to the phantom to simulate image acquisition.

[0427] Control the following parameters to construct a digital phantom.

[0428] Image size:

[0429] Image size: 10 x 10 mm (24 x 24 pixels)

[0430] Pixel size = 0.41 x 0.41 mm

[0431] Blood vessel size:

[0432] Lumen radius = 1.5 mm

[0433] Outer wall thickness = 0.5 mm

[0434] PVAT thickness = 2.0 mm

[0435] Image intensity:

[0436] Lumen = 190 HU

[0437] Outer wall = -30 HU

[0438] PVAT = -75 HU

[0439] The gradual increase in focal inflammation:

[0440] -75 to -49 HU, in increments of 2 HU.

[0441] The attenuation of PVAT was quantified as the average attenuation of all voxels within the entire circumference of the PVAT region around the coronary artery. Figure 6 c) represents the conventional method, as well as the radial segmental method, in which the attenuation of PVAT is quantified as the average attenuation of voxels within a quarter radial segment of the PVAT region around the coronary artery, according to the present invention. Figure 6 d)). Multiple simulations were performed using different instances of Gaussian noise at each inflammation intensity.

[0442] As inflammation increased from -75 HU to -49 HU, the measured PVAT decay change (e.g. Figure 7 As shown, a comparison was made between two measurement techniques (radial segment and full circumference).

[0443] The PVAT attenuation results of 50 simulations for each inflammation intensity are as follows: Figure 8 As shown, the results of 500 simulations at each inflammation intensity are as follows: Figure 9 As shown. The interquartile range of the radial segment method becomes distinguishable from the baseline (i.e., non-inflammatory vessels) at a lower inflammatory intensity (-69 HU) than the full circumferential method (black box; -63 HU).

[0444] Figure 8 The horizontal dashed line in the chart shows the upper quarter distance of the baseline using the radial segment method, while the horizontal dotted line shows the equivalent upper quarter distance using the full-circle method.

[0445] As can be seen from the graphs, the radial segmental method according to the present invention can detect inflammatory areas at lower intensities than conventional methods. Therefore, the method of the present invention helps to detect PVAT inflammation earlier and enable earlier medical intervention.

Claims

1. A method for characterizing the pathology or condition of a vascular region in a subject, the method comprising: (i) Provide medical imaging data acquired along the length of the blood vessel; as well as (i) Determine a radial segment of the blood vessel extending from a point at the center of the blood vessel to a distance extending beyond the outer wall of the blood vessel to the area surrounding the blood vessel; (ii) Determine the values ​​of the radioactivity density and / or one or more radiomics characteristics of the radial segment; as well as (ii) Compare each value determined in (ii) with a predetermined value, or use an absolute value, to generate an output value indicating the pathology or condition of the vascular region; At least one of the method steps is implemented by a computer.

2. The method according to claim 1, wherein, Step (ii) includes determining the radiometric density of a voxel corresponding to the perivascular tissue in the perivascular region of each radial segment; preferably, step (ii) includes determining the radiometric density of perivascular adipose tissue (PVAT).

3. The method according to any of the preceding claims, wherein, The distance extending beyond the outer wall of the blood vessel is: (i) A standard distance that is not equal to or related to the diameter or radius of the underlying vessel; (ii) is a derived value, multiple of, or equal to the radius or diameter of the basic vessel; or (iii) 0.1 mm to 3 cm, preferably 0.2 mm to 2.5 cm, more preferably 0.3 mm to 2.25 cm, and even more preferably 0.4 mm to 2 cm.

4. The method according to any of the preceding claims, wherein, The output value determined in (b) provides a measure of the texture of the vascular region.

5. The method according to any of the preceding claims, wherein, The output value determined in step (b) provides a measure of inflammation in the vascular region.

6. The method according to any of the preceding claims, wherein, The method is used to identify subclinical plaques, preplaque states, and / or early atherosclerotic disease activity in the vessel walls of inflammatory vascular regions.

7. The method according to any of the preceding claims, further comprising: The steps for identifying inflammatory plaques within the vascular region.

8. The method according to any of the preceding claims, further comprising: The step of identifying the inflammatory portion of plaque within the vascular region.

9. The method of claim 8, wherein the calcified areas of the plaque are not classified as inflammatory.

10. The method according to any one of claims 6 to 9, further comprising: The steps of classifying the patches into low-risk, medium-risk, high-risk, or very high-risk categories.

11. A method for characterizing plaque in the blood vessels of a subject, the method comprising: (a) Provide medical imaging data acquired along the length of the blood vessel; as well as (i) Identify voxels corresponding to plaques by their radioactivity; (ii) Determine a radial segment of the blood vessel containing a voxel corresponding to the plaque, wherein the radial segment extends from a point at the center of the blood vessel to a distance extending beyond the outer wall of the blood vessel; (iii) Determine the values ​​of the radioactivity density and / or one or more radiomics characteristics of the radial segment; (b) Compare each value determined in (iii) with a predetermined threshold, or use the absolute value determined in (iii), to generate an output value indicating inflammation of the plaque; At least one of the method steps is implemented by a computer.

12. The method according to claim 11, wherein, Step (ii) includes determining the radiometric density of a voxel corresponding to perivascular tissue in the radial segment, preferably wherein step (ii) includes determining the radiometric density of a voxel corresponding to perivascular adipose tissue (PVAT).

13. The method according to claim 11 or 12, wherein, The radial segment is: (i) Predefined geometry; (ii) The radius of the radial segment is defined by the physical limit of the voxel corresponding to the patch.

14. The method according to any one of claims 11 to 13, wherein, The therapy was administered to the subjects after the inflammatory plaques were identified.

15. The method according to any one of claims 11 to 14, wherein, The plaque is characterized as inflammatory, and the inflammatory plaque is either a fibrous plaque or a non-fibrous plaque.

16. The method according to any one of claims 11 to 15, wherein, The inflammatory plaques were classified as hotspots, unstable, high-risk, or potential culprit lesions.

17. The method according to any of the preceding claims, wherein, The medical imaging data is computed tomography (CT) data.

18. The method according to any preceding claim, comprising: Determine the values ​​of at least two radiomics features; The value of each radiomics feature is included in step (b).

19. The method according to any of the preceding claims, wherein, The vascular region includes: the entire radial segment, the perivascular portion of the radial segment, or the vascular portion of the radial segment.

20. The method according to any of the preceding claims, wherein, The radiomics features are selected from the radiomics features of clusters 1 to 9, wherein: Cluster 1 consists of short-run high gray level emphasis, high gray level emphasis, high gray level run emphasis, autocorrelation, average, joint average and high gray level region emphasis; Cluster 2 consists of skewness, skewness LLL, kurtosis, 90th percentile, 90th percentile LLL, median LLL, kurtosis LLL, and median; Cluster 3 consists of run entropy, dependency entropy LLL, dependency entropy, region entropy LLL, run entropy LLL, and average value LLL; Cluster 4 consists of small-area low-gray-level emphasis, low-gray-level area emphasis, short-run low-gray-level emphasis, low-gray-level run emphasis, low-gray-level emphasis, low-dependency low-gray-level emphasis, gray-level variance LLL (GLSZM), gray-level variance (GLDM), variance, gray-level variance (GLDM), difference variance LLL, gray-level variance LLL (GLRLM), variance LLL, gray-level variance LLL (GLDM), sum of squares, contrast LLL, mean absolute deviation, interquartile range, robust mean absolute deviation, long-run low-gray-level emphasis, difference variance, gray-level variance (GLSZM), normalized inverse difference moment, mean absolute deviation LLL, sum of squares LLL, and contrast. Cluster 5 consists of regional entropy, normalized gray level nonuniformity (GLRLM), normalized gray level nonuniformity LLL (GLRLM), and entropy, joint energy, entropy, normalized gray level nonuniformity (GLDM), joint energy, normalized gray level nonuniformity LLL (GLDM), uniformity LLL, and entropy LLL and uniformity. Cluster 6 consists of regional entropy HHH, normalized size regional non-uniformity HHH, and small region emphasis HHH; Cluster 7 consists of strength, roughness HLL, roughness, roughness LHL, roughness LLL, roughness LLH, roughness HHH, roughness HLH, roughness HHL and roughness LHH; Cluster 8 consists of cluster tendency (LLL), cluster tendency, sum of squares (LLL), mean absolute deviation (LLL), gray level variance (GLDM), variance (LLL), gray level variance (GLRLM), gray level variance (GLRLM), robust mean absolute deviation (LLL), gray level variance (GLDM), variance, mean absolute deviation, cluster salientity, sum of entropy (LLL), interquartile range (LLL), gray level variance (GLSZM), sum of squares, robust mean absolute deviation, sum of entropy, interquartile range, cluster salientity (LLL), entropy (LLL), 10th percentile (LLL), 10th percentile; and Cluster 9 consists of size region nonuniformity LLL, dependent nonuniformity HLL, gray level nonuniformity HLL (GLSZM), gray level nonuniformity (GLSZM), run length nonuniformity HHL, run length nonuniformity LHL, dependent nonuniformity LHL, dependent nonuniformity, run length nonuniformity HLH, busyness, run length nonuniformity LLH, dependent nonuniformity LLH, dependent nonuniformity LLL, size region nonuniformity, energy HLL, run length nonuniformity LHH, size region nonuniformity HLL, gray level nonuniformity LLH (GLSZM), gray level nonuniformity LHL (GLSZM), gray level nonuniformity LLL (GLSZM), run length nonuniformity HLL, gray level nonuniformity HLH (GLSZM), gray level nonuniformity HHL (GLSZM), run length nonuniformity and run length nonuniformity HHH. Preferably, at least two radiomics features are selected; preferably, the at least two radiomics features are each selected from different clusters.

21. The method according to any of the preceding claims, wherein, The radial segments have a predefined geometry; preferably, the angle of the segment from a point in the center of the blood vessel is 1 to 359°, more preferably 2 to 270°, more preferably 10 to 180°, and even more preferably 30 to 150°.

22. The method according to any of the preceding claims, wherein, The data were acquired from computed tomography scans along the length of the right coronary artery, left anterior descending branch, left circumflex branch, aorta, carotid artery, or femoral artery.

23. The method according to any of the preceding claims, wherein, The threshold is the distribution of attenuation values, preferably wherein the threshold is vascular-specific and / or specific to how the attenuation values ​​are derived along the proximal or distal end of the blood vessel.

24. The method according to any of the preceding claims, wherein, The predetermined threshold is adjusted for biological factors, preferably selected from one or more of the following: the subject's age, the subject's sex, the subject's ethnicity, background adipocyte size, partial volume effect, and vascular type.

25. The method according to any of the preceding claims, wherein, The predetermined threshold is adjusted for technical factors, preferably, wherein the technical factors are selected from one or more of the following: tube voltage of the CT scanner, reconstruction algorithm, iodinated contrast agent, contrast agent type, injection rate, aortic enhancement degree, left ventricular blood pool enhancement degree, signal-to-noise ratio, contrast-to-noise ratio, milliampere, cardiac gating method, single-energy and multi-energy image acquisition, CT scanner type, heart rate, heart rhythm or blood pressure.

26. The method according to any of the preceding claims, wherein, The method is implemented by a computer.

27. The method according to any of the preceding claims, wherein, The method is used to monitor the progression of plaques.

28. The method according to any of the preceding claims, wherein, The method is used to predict the risk of cardiac death or the risk of a subject experiencing a cardiovascular event.

29. The method according to claim 20, wherein, The output value is combined with other biomarkers to predict the risk of cardiac death or the risk of a subject experiencing a cardiovascular event.

30. The method according to any of the preceding claims, wherein, The method is used to guide drug treatment decisions and / or monitor responses to medical care.

31. The method according to any of the preceding claims, wherein, The method is used for non-invasive monitoring of aortic aneurysms or carotid plaques.

32. The method according to any of the preceding claims, wherein, The method is used to stratify subjects based on their risk of cardiac death or the risk of experiencing cardiovascular events.

33. The method according to any of the preceding claims, wherein, The method is used to quantify one or more of the following: vascular inflammation, fibrosis, edema, and vascular distribution.

34. The method according to any of the preceding claims, wherein, The output values ​​are compared with earlier scans of the same subject to calculate local changes in the inflammatory state.

35. A computer program product comprising executable instructions, which, when executed by one or more processors, cause the one or more processors to perform the method of any of the preceding claims.

36. A system comprising one or more processors configured to perform the method of any of the preceding claims.

Images