Vulnerable plaque detection method and related apparatus
By constructing radiomics models and fusion prediction models, and combining HRMR-VWI image data and clinical features, the problem of accurately identifying vulnerable plaques in vascular wall images was solved, improving the accuracy of plaque detection and the reliability of stroke risk assessment.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- TONGJI HOSPITAL ATTACHED TO TONGJI MEDICAL COLLEGE HUAZHONG SCI TECH
- Filing Date
- 2022-09-08
- Publication Date
- 2026-05-19
AI Technical Summary
Existing technologies struggle to accurately identify vulnerable plaques in vascular wall images, especially intracranial plaques, making it difficult to improve stroke prognosis and prevent recurrence.
By constructing a radiomics model based on plaque features from at least two experimental samples, and combining the traditional model with the radiomics model to determine a fusion prediction model, a method for assessing vulnerable plaques is established by using high-resolution magnetic resonance imaging (HRMR-VWI) image data to extract plaque features and risk factors.
It enables accurate assessment of vulnerable plaques, improves the accuracy and sensitivity of plaque detection, and enhances the reliability of stroke risk assessment.
Smart Images
Figure CN115511797B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of medical identification, and in particular to a method and related equipment for detecting vulnerable plaques. Background Technology
[0002] Atherosclerosis is the most common cause of stroke. Vulnerable plaques are those that cause stroke events. Stroke patients may have multiple plaques in their intracranial blood vessels. Therefore, finding and treating vulnerable plaques is crucial for improving prognosis and preventing stroke recurrence.
[0003] The fatty components and fibrous caps that we observe in carotid artery plaques are difficult to observe in intracranial plaques. Generally speaking, HRMR-VWI can visualize carotid artery plaques very well. However, for intracranial plaques, due to the small size of the blood vessels and the influence of clinical diseases, not all plaques are typical and easy to identify. Accurately and clearly displaying all plaques and making a definitive diagnosis still presents certain difficulties, and identifying vulnerable plaques is even more challenging. Summary of the Invention
[0004] In view of the above problems, the present invention provides a method and related equipment for detecting vulnerable plaques, the main purpose of which is to solve the problem of the lack of an accurate method for identifying vulnerable plaques in blood vessel wall images.
[0005] To address at least one of the aforementioned technical problems, in a first aspect, the present invention provides a method for detecting vulnerable plaques, the method comprising:
[0006] A radiomics model was constructed using plain scan plus enhanced 3D HRMR-VWI images based on patch features from at least two experimental samples.
[0007] A fusion prediction model was determined based on traditional models and the aforementioned radiomics models.
[0008] Based on the above fusion prediction model, vulnerable plaques in the target sample are identified.
[0009] Optionally, the above methods also include:
[0010] Acquire image data for at least two test samples, including high-resolution MRI images before and after enhancement.
[0011] Optionally, the above methods also include:
[0012] Based on the above image data, patch information is determined.
[0013] The plaque information includes: plaque diameter, minimum luminal area, intraplaque hemorrhage, minimum luminal diameter, stenosis rate, plaque burden, enhancement rate, and remodeling index.
[0014] Optionally, the above methods also include:
[0015] A conventional model is constructed based on the plaque characteristics and risk factors of at least two test samples, wherein the aforementioned risk factors are used to characterize factors that are likely to lead to the formation of vulnerable plaques.
[0016] Optionally, the vulnerable plaques of the target sample determined based on the above-mentioned fusion prediction model include:
[0017] Based on the above-mentioned fusion prediction model and the risk factors of the target sample, the vulnerable plaques of the target sample are identified.
[0018] Optionally, the determination of vulnerable plaques in the target sample based on the aforementioned fusion prediction model and the aforementioned risk factors of the target sample includes:
[0019] Obtain the patch characteristics and risk factors of the target sample;
[0020] Based on the patch characteristics and risk factors of the target samples, the assessment parameters are determined using the aforementioned fusion prediction model.
[0021] Vulnerable plaques were identified based on the above evaluation parameters.
[0022] Optionally, the vulnerable plaques of the target sample determined based on the above-mentioned fusion prediction model include:
[0023] Patches that meet the preset conditions for the above evaluation parameters are identified as the aforementioned vulnerable patches.
[0024] The evaluation parameters mentioned above include: AUC, sensitivity, specificity, accuracy, negative predictive value, and positive predictive value.
[0025] Secondly, embodiments of the present invention also provide a vulnerable plaque detection device, comprising:
[0026] The building blocks are used to construct radiomics models based on patch features from at least two experimental samples using plain scan plus enhanced 3D HRMR-VWI images;
[0027] The unit is determined, and a fusion prediction model is determined based on the traditional model and the above-mentioned radiomics model;
[0028] The second determining unit is used by the aforementioned fusion prediction model to determine vulnerable patches of the target sample.
[0029] To achieve the above objectives, according to a third aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium comprising a stored program, wherein, when the program is executed by a processor, the steps of the above-described vulnerable plaque detection method are implemented.
[0030] To achieve the above objectives, according to a fourth aspect of the present invention, an electronic device is provided, comprising at least one processor and at least one memory connected to the processor; wherein the processor is configured to invoke program instructions in the memory to execute the steps of the above-described vulnerable plaque detection method.
[0031] By employing the above technical solutions, the vulnerable plaque detection method and related equipment provided by this invention address the current lack of an accurate method for identifying vulnerable plaques in vascular wall images. This invention constructs a radiomics model based on plaque features from at least two experimental samples using plain and enhanced 3D HRMR-VWI images; determines a fusion prediction model based on a traditional model and the aforementioned radiomics model; and identifies vulnerable plaques in the target sample based on the fusion prediction model. In this approach, plaques are segmented, and a large number of image features are extracted from the images in a high-throughput manner, enabling the detection of massive amounts of information invisible to the human eye. Based on 3D sequences, using radiomics methods, the traditional model and the aforementioned radiomics model are fused on pre- and post-enhanced 3D HRMR-VWI images to assess the differences between vulnerable and non-vulnerable plaques in the middle cerebral artery, thereby providing a more accurate assessment of vulnerable plaques.
[0032] Accordingly, the vulnerable plaque detection device, equipment, and computer-readable storage medium provided in the embodiments of the present invention also have the above-mentioned technical effects.
[0033] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and in order to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description
[0034] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:
[0035] Figure 1 A schematic flowchart of a vulnerable plaque detection method provided by an embodiment of the present invention is shown;
[0036] Figure 2 This diagram illustrates the composition of a vulnerable plaque detection device according to an embodiment of the present invention.
[0037] Figure 3 This diagram illustrates the composition of an electronic device for detecting vulnerable plaques according to an embodiment of the present invention. Detailed Implementation
[0038] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this invention will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.
[0039] To address the current lack of an accurate method for identifying vulnerable plaques in blood vessel wall images, this invention provides a vulnerable plaque detection method, such as... Figure 1 As shown, the method includes:
[0040] S101. Construct a radiomics model based on patch features from at least two experimental samples using plain scan plus enhanced 3D HRMR-VWI images;
[0041] For example, in the ITK Snap software, the patch is delineated, and the 3D HRMR-VWI images before and after enhancement are imported into ITKSNAP to delineate and extract the ROI of the entire patch.
[0042] For example, the XGB model is an emerging modeling method. The core idea of XGB is to use the classification and regression tree (CART) as the basic classifier, and reduce the residual loss between the true value and the predicted value through iterative processing. The optimization is carried out along the gradient direction with the goal of reducing the residual loss, thereby continuously splitting features to grow the tree.
[0043] For example, this method uses PyRadiomics to extract features from the ROIs obtained from the 3D HRMR-VWI before and after enhancement, respectively. Maximum correlation minimum redundancy radiomics feature selection is used, leaving 10 features for each sequence. An unenhanced + enhanced 3D HRMR-VWI model is then established using extreme gradient boosting (XGB).
[0044] S102. Determine the fusion prediction model based on the traditional model and the above-mentioned radiomics model;
[0045] For example, traditional models are constructed from the above plaque characteristics and risk factors.
[0046] For example, the XGB model is an emerging modeling method. The core idea of XGB is to use the classification and regression tree (CART) as the basic classifier, and reduce the residual loss between the true value and the predicted value through iterative processing. The optimization is carried out along the gradient direction with the goal of reducing the residual loss, thereby continuously splitting features to grow the tree.
[0047] For example, indicators selected from clinical features and 3D HRMR-VWI features before and after enhancement are used to construct the final fusion prediction model.
[0048] S103. Based on the above fusion prediction model, determine the vulnerable plaques of the target sample.
[0049] For example, indicators selected from clinical features and 3D HRMR-VWI features before and after enhancement were used to construct the final fusion prediction model. The predictive performance of the model was evaluated using receiver operating characteristic (ROC). The area under the ROC curve (AUC), sensitivity, specificity, accuracy, negative predictive value (NPV), and positive predictive value (PPV) were calculated, and the reliability of the model was evaluated using a calibration curve. Furthermore, the clinical utility of our established nomota was determined by quantifying the net benefit at different threshold probabilities using a clinical decision curve.
[0050] For example, compared to 2D-based images that only measure the histogram information of the largest slice, this method also has better predictive performance when studying the entire patch on 3D HRMR-VWI.
[0051] For example, this method combines clinical information and radiomics to establish a fusion prediction model for vulnerable plaques in the middle cerebral artery on 3D HRMR-VWI, achieving excellent performance. In traditional information, only intraplaque hemorrhage is an independent predictor of vulnerable plaques. Radiomics prediction models outperform clinical models. The fusion prediction model of this method provides a new approach for identifying vulnerable plaques in vessel wall images.
[0052] By employing the above technical solution, the vulnerable plaque detection method provided by this invention addresses the current lack of an accurate method for identifying vulnerable plaques in vascular wall images. This invention constructs a radiomics model based on plaque features from at least two experimental samples using plain and enhanced 3D HRMR-VWI images; determines a fusion prediction model based on a traditional model and the aforementioned radiomics model; and identifies vulnerable plaques in the target sample based on the fusion prediction model. In this approach, plaques are segmented, and a large number of image features are extracted from the images in a high-throughput manner, enabling the detection of massive amounts of information invisible to the human eye. Based on 3D sequences, using radiomics methods, the traditional model and the aforementioned radiomics model are fused on pre- and post-enhanced 3D HRMR-VWI images to assess the differences between vulnerable and non-vulnerable plaques in the middle cerebral artery, thereby providing a more accurate assessment of vulnerable plaques.
[0053] In one embodiment, the above method further includes:
[0054] Acquire image data for at least two test samples, including high-resolution MRI images before and after enhancement.
[0055] For example, the test samples were subjected to magnetic resonance imaging (MRI) to obtain image data of the test samples to identify several atherosclerotic plaques in the middle cerebral arteries. The samples were randomly divided into training and validation sets in a 7:3 ratio. The MRI examination included high-resolution magnetic resonance imaging (3D HRMR-VWI) before and after enhancement. Based on the MRI images and clinical symptoms, the plaques were classified as vulnerable plaques and non-vulnerable plaques.
[0056] In one embodiment, the above method further includes:
[0057] Based on the above image data, patch information is determined.
[0058] The plaque information includes: plaque diameter, minimum luminal area, intraplaque hemorrhage, minimum luminal diameter, stenosis rate, plaque burden, enhancement rate, and remodeling index.
[0059] For example, 3D HRMR-VWI and TOF-MRA are imported into the United Imaging Post-Processing Workstation (Uws-MR) to measure the plaque diameter (DP), vessel diameter (DV), luminal area (SL), vessel area (SV), plain scan plaque signal, plain scan surrounding gray matter signal, and enhanced plaque signal and enhanced surrounding gray matter signal, as well as the distal normal vessel area. The plaque is delineated in the software, and the 3D HRMR-VWI images before and after enhancement are imported into ITK SNAP to delineate the entire plaque's ROI.
[0060] For example, plaque characteristics are determined based on the data measured and acquired above: plaque diameter, minimum lumen area, presence of intraplaque hemorrhage, minimum lumen diameter, stenosis rate, plaque burden, enhancement rate, and remodeling index.
[0061] Stenosis rate:
[0062]
[0063] Where Ds is the diameter at the narrowest point of the lumen, Dd is the diameter of the distal normal blood vessel, and Dp is the diameter of the proximal normal blood vessel.
[0064] Plaque load:
[0065]
[0066] Reshaping Index:
[0067]
[0068] Enhancement rate:
[0069]
[0070] In one embodiment, the above method further includes:
[0071] A conventional model is constructed based on the plaque characteristics and risk factors of at least two test samples, wherein the aforementioned risk factors are used to characterize factors that are likely to lead to the formation of vulnerable plaques.
[0072] An anti-scan 3D HRMR-VWI model and an enhanced 3D HRMR-VWI model were established using extreme gradient boosting (XGB).
[0073] For example, traditional models assessing the effectiveness of vulnerable plaques, through univariate evaluation, reveal differences in minimum luminal area, intraplaque hemorrhage, minimum luminal diameter, stenosis rate, plaque burden, and enhancement rate between the two groups (training and validation sets). Vulnerable plaques have smaller minimum luminal area and minimum luminal diameter than non-vulnerable plaques, higher stenosis rates, higher intraplaque hemorrhage rates, higher plaque burden, and higher enhancement rates. Multivariate logistic regression analysis confirms that plaques with only intraplaque hemorrhage are significant in identifying vulnerable plaques.
[0074] In one embodiment, the determination of vulnerable plaques in the target sample based on the aforementioned fusion prediction model includes:
[0075] Based on the above-mentioned fusion prediction model and the risk factors of the target sample, the vulnerable plaques of the target sample are identified.
[0076] For example, the test sample consists of several images of the blood vessel wall. Risk factors for the test sample include: the subject's gender, age, hypertension, hyperlipidemia, diabetes, smoking, alcohol consumption, history of coronary heart disease, and history of stroke. Traditional models are constructed based on these plaque characteristics and risk factors.
[0077] In one embodiment, the determination of vulnerable plaques in the target sample based on the aforementioned fusion prediction model and the aforementioned risk factors of the target sample includes:
[0078] Obtain the patch characteristics and risk factors of the target sample;
[0079] Based on the patch characteristics and risk factors of the target samples, the assessment parameters are determined using the aforementioned fusion prediction model.
[0080] Vulnerable plaques were identified based on the above evaluation parameters.
[0081] For example, after determining the fusion prediction model and applying it, firstly, the patch characteristics and risk factors of the target sample are obtained, and then the patch characteristics and risk factors of the target sample are substituted into the fusion prediction model to determine the evaluation parameters. Then, based on the evaluation parameters, the patches of the target sample are evaluated to determine vulnerable patches.
[0082] In one embodiment, the determination of vulnerable plaques in the target sample based on the aforementioned fusion prediction model includes:
[0083] Patches that meet the preset conditions for the above evaluation parameters are identified as the aforementioned vulnerable patches.
[0084] The evaluation parameters mentioned above include: AUC, sensitivity, specificity, accuracy, negative predictive value, and positive predictive value.
[0085] For example, based on the following plaque assessment parameters: AUC, sensitivity, specificity, accuracy, negative predictive value, and positive predictive value, it is possible to determine whether vulnerable plaques exist and which plaques are vulnerable.
[0086] For example, in order to identify vulnerable plaques in symptomatic middle cerebral arteries, this study combined traditional information (risk factors for stroke and plaque information) with radiomics information from 3D HRMR-VWI before and after enhancement, and established a predictive model using radiomics methods. The results of this method are as follows: the AUC for detecting vulnerable plaques was 0.949, and its sensitivity and specificity were 91.67% and 83.67%, respectively, achieving very high efficacy.
[0087] Furthermore, the specific implementation process of the embodiments of the present invention is shown below:
[0088] Based on the patch features of at least two experimental samples, a radiomics model was constructed using plain scan plus enhanced 3D HRMR-VWI images; patches were delineated in ITK SNAP software, and the ROI of the entire patch was delineated and extracted by importing the 3D HRMR-VWI images before and after enhancement into ITK SNAP.
[0089] For example, the XGB model is an emerging modeling method. The core idea of XGB is to use the classification and regression tree (CART) as the basic classifier, and reduce the residual loss between the true value and the predicted value through iterative processing. The optimization is carried out along the gradient direction with the goal of reducing the residual loss, thereby continuously splitting features to grow the tree.
[0090] For example, features were extracted from the ROIs obtained from the 3D HRMR-VWI before and after enhancement using PyRadiomics. Maximum correlation minimum redundancy radiomics feature selection was employed, leaving 10 features for each sequence. An unenhanced + enhanced 3D HRMR-VWI model was then established using extreme gradient boosting (XGB).
[0091] A fusion prediction model was determined based on the traditional model and the aforementioned radiomics model. The traditional model was constructed from the above-mentioned plaque characteristics and risk factors.
[0092] For example, 3D HRMR-VWI and TOF-MRA are imported into the United Imaging Post-Processing Workstation (Uws-MR) to measure the plaque diameter DP, vessel diameter DV, lumen area SL, vessel area SV, plain scan plaque signal, plain scan surrounding gray matter signal, and enhanced plaque signal and enhanced surrounding gray matter signal, as well as the area of the distal normal vessel.
[0093] For example, plaque characteristics are determined based on measured and acquired data: plaque diameter, minimum lumen area, presence of intraplaque hemorrhage, minimum lumen diameter, stenosis rate, plaque burden, enhancement rate, and remodeling index.
[0094] A conventional model was constructed based on plaque characteristics and risk factors from at least two experimental samples. This model was used to evaluate the effectiveness of vulnerable plaques. Through univariate evaluation, differences were found between the two groups (training set and validation set) in minimum lumen area, intraplaque hemorrhage, minimum lumen diameter, stenosis rate, plaque burden, and enhancement rate. Vulnerable plaques had smaller minimum lumen area and minimum lumen diameter than non-vulnerable plaques. Vulnerable plaques had a higher stenosis rate than non-vulnerable plaques. Intraplaque hemorrhage was higher in vulnerable plaques than in non-vulnerable plaques. The plaque burden and enhancement rate of vulnerable plaques were also higher in vulnerable plaques than in non-vulnerable plaques.
[0095] Vulnerable plaques in the target sample are identified based on the fusion prediction model and the risk factors of the target sample.
[0096] For example, indicators selected from clinical features and 3D HRMR-VWI features before and after enhancement are used to construct the final fusion prediction model.
[0097] Vulnerable plaques in the target sample are identified based on a fusion prediction model.
[0098] For example, indicators selected from clinical features and pre- and post-enhancement 3D HRMR-VWI features were used to construct the final fusion prediction model. The predictive performance of the model was evaluated using receiver operating characteristic (ROC) curves. The area under the ROC curve (AUC), sensitivity, specificity, accuracy, negative predictive value (NPV), and positive predictive value (PPV) were calculated, and the reliability of the model was evaluated using calibration curves. Furthermore, the clinical utility of our established nomograph was determined by quantifying the net benefit at different threshold probabilities using clinical decision curves. Compared to 2D-based images, which only measure histogram information at the largest slice, the study showed better predictive efficacy by measuring the entire plaque on 3D HRMR-VWI.
[0099] Furthermore, as a response to the above Figure 1 In addition to the implementation of the method shown, this embodiment of the invention also provides a vulnerable plaque detection device for detecting the aforementioned plaques. Figure 1 The method shown is implemented accordingly. This device embodiment corresponds to the foregoing method embodiment. For ease of reading, this device embodiment will not repeat the details of the foregoing method embodiment, but it should be clear that the device in this embodiment can implement all the contents of the foregoing method embodiment. Figure 2 As shown, the device includes: a construction unit 21, a determination unit 22, and a second determination unit 23, wherein...
[0100] Building unit 21 is used to construct a radiomics model based on patch features from at least two experimental samples using plain scan plus enhanced 3D HRMR-VWI images;
[0101] Unit 22 determines the fusion prediction model based on the traditional model and the above-mentioned radiomics model;
[0102] The second determining unit 23 is used to determine the vulnerable patches of the target sample by the above-mentioned fusion prediction model.
[0103] For example, the above-mentioned unit is also used for:
[0104] Acquire image data for at least two test samples, including high-resolution MRI images before and after enhancement.
[0105] For example, the above-mentioned unit is also used for:
[0106] Based on the above image data, patch information is determined.
[0107] The plaque information includes: plaque diameter, minimum luminal area, intraplaque hemorrhage, minimum luminal diameter, stenosis rate, plaque burden, enhancement rate, and remodeling index.
[0108] For example, the above-mentioned unit is also used for:
[0109] A conventional model is constructed based on the plaque characteristics and risk factors of at least two test samples, wherein the aforementioned risk factors are used to characterize factors that are likely to lead to the formation of vulnerable plaques.
[0110] For example, the above-mentioned determination of vulnerable patches of the target sample based on the above-mentioned fusion prediction model includes:
[0111] Based on the above-mentioned fusion prediction model and the risk factors of the target sample, the vulnerable plaques of the target sample are identified.
[0112] For example, the above-mentioned determination of vulnerable plaques in the target sample based on the above-mentioned fusion prediction model and the above-mentioned risk factors of the target sample includes:
[0113] Obtain the patch characteristics and risk factors of the target sample;
[0114] Based on the patch characteristics and risk factors of the target samples, the assessment parameters are determined using the aforementioned fusion prediction model.
[0115] Vulnerable plaques were identified based on the above evaluation parameters.
[0116] For example, the above-mentioned determination of vulnerable patches of the target sample based on the above-mentioned fusion prediction model includes:
[0117] Patches that meet the preset conditions for the above evaluation parameters are identified as the aforementioned vulnerable patches.
[0118] The evaluation parameters mentioned above include: AUC, sensitivity, specificity, accuracy, negative predictive value, and positive predictive value.
[0119] By employing the above technical solution, the vulnerable plaque detection device provided by this invention addresses the current lack of a method for accurately identifying vulnerable plaques in vascular wall images. This invention constructs a radiomics model based on plaque features from at least two experimental samples using plain and enhanced 3D HRMR-VWI images; determines a fusion prediction model based on a traditional model and the aforementioned radiomics model; and identifies vulnerable plaques in the target sample based on the fusion prediction model. In this approach, plaques are segmented, and a large number of image features are extracted from the images in high throughput, enabling the detection of massive amounts of information invisible to the human eye. Based on 3D sequences, using radiomics methods, the traditional model and the aforementioned radiomics model are fused on pre- and post-enhanced 3D HRMR-VWI images to assess the differences between vulnerable and non-vulnerable plaques in the middle cerebral artery, thereby providing a more accurate assessment of vulnerable plaques.
[0120] The processor contains a kernel, which retrieves the corresponding program units from memory. One or more kernels can be configured, and by adjusting kernel parameters, a vulnerable plaque detection method can be implemented, addressing the current lack of an accurate method for identifying vulnerable plaques in blood vessel wall images.
[0121] This invention provides a computer-readable storage medium including a stored program that, when executed by a processor, implements the aforementioned vulnerable plaque detection method.
[0122] This invention provides a processor for running a program, wherein the program executes the aforementioned vulnerable plaque detection method.
[0123] This invention provides an electronic device, which includes at least one processor and at least one memory connected to the processor; wherein the processor is used to call program instructions in the memory to execute the vulnerable plaque detection method described above.
[0124] This invention provides an electronic device 30, such as... Figure 3 As shown, the electronic device includes at least one processor 301, and at least one memory 302 and bus 303 connected to the processor; wherein, the processor 301 and the memory 302 communicate with each other through the bus 303; the processor 301 is used to call program instructions in the memory to execute the above-mentioned vulnerable plaque detection method.
[0125] The smart electronic devices mentioned in this article can be PCs, tablets, mobile phones, etc.
[0126] This application also provides a computer program product, which, when executed on a process management electronic device, is suitable for executing a program that initializes the following method steps:
[0127] A radiomics model was constructed using plain scan plus enhanced 3D HRMR-VWI images based on patch features from at least two experimental samples.
[0128] A fusion prediction model was determined based on traditional models and the aforementioned radiomics models.
[0129] Based on the above fusion prediction model, vulnerable plaques in the target sample are identified.
[0130] Furthermore, the above methods also include:
[0131] Acquire image data for at least two test samples, including high-resolution MRI images before and after enhancement.
[0132] Furthermore, the above methods also include:
[0133] Based on the above image data, patch information is determined.
[0134] The plaque information includes: plaque diameter, minimum luminal area, intraplaque hemorrhage, minimum luminal diameter, stenosis rate, plaque burden, enhancement rate, and remodeling index.
[0135] Furthermore, the above methods also include:
[0136] A conventional model is constructed based on the plaque characteristics and risk factors of at least two test samples, wherein the aforementioned risk factors are used to characterize factors that are likely to lead to the formation of vulnerable plaques.
[0137] Furthermore, the vulnerable plaques of the target sample determined based on the aforementioned fusion prediction model include:
[0138] Based on the above-mentioned fusion prediction model and the risk factors of the target sample, the vulnerable plaques of the target sample are identified.
[0139] Furthermore, the identification of vulnerable plaques in the target sample based on the aforementioned fusion prediction model and the aforementioned risk factors of the target sample includes:
[0140] Obtain the patch characteristics and risk factors of the target sample;
[0141] Based on the patch characteristics and risk factors of the target samples, the assessment parameters are determined using the aforementioned fusion prediction model.
[0142] Vulnerable plaques were identified based on the above evaluation parameters.
[0143] Furthermore, the vulnerable plaques of the target sample determined based on the aforementioned fusion prediction model include:
[0144] Patches that meet the preset conditions for the above evaluation parameters are identified as the aforementioned vulnerable patches.
[0145] The evaluation parameters mentioned above include: AUC, sensitivity, specificity, accuracy, negative predictive value, and positive predictive value.
[0146] This application is described with reference to flowchart illustrations and / or block diagrams of methods, electronic devices (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable process management electronic device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable process management electronic device, generate instructions for implementing the process... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0147] In a typical configuration, an electronic device includes one or more processors (CPUs), memory, and a bus. The electronic device may also include input / output interfaces, network interfaces, etc.
[0148] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, like read-only memory (ROM) or flash RAM, and memory includes at least one memory chip. Memory is an example of computer-readable media.
[0149] Computer-readable media, including both permanent and non-permanent, removable and non-removable media, can store information using any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer-readable storage media for computers include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage electronic devices, or any other non-transferable medium that can be used to store information accessible to a computing electronic device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.
[0150] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or electronic device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or electronic device. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or electronic device that includes that element.
[0151] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable, computer-readable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0152] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.
Claims
1. A method for detecting vulnerable plaques, characterized in that, include: Based on the patch features of at least two experimental samples, a radiomics model was constructed using plain scan plus enhanced 3D HRMR-VWI images; patches were delineated in ITK SNAP software, and the 3D HRMR-VWI images before and after enhancement were imported into ITK SNAP to delineate and extract the ROI of the entire patch; An XGB model is constructed, which uses classification and regression trees as the basic classifier. The residual loss between the true value and the predicted value is reduced through iterative processing. The model is optimized along the gradient direction with the goal of reducing the residual loss, thereby continuously splitting features to grow the tree. Features were extracted from the ROIs obtained from the 3D HRMR-VWI before and after enhancement using PyRadiomics. The maximum correlation minimum redundancy radiomics feature selection was used, leaving 10 features for each sequence. The extreme gradient boosting method was used to build a plain scan + enhancement 3D HRMR-VWI model. A fusion prediction model is determined based on the traditional model and the aforementioned radiomics model; the traditional model is constructed from the above-mentioned patch characteristics and risk factors. The indicators selected from clinical features and the 3D HRMR-VWI features before and after enhancement were used to construct the final fusion prediction model; Based on the fusion prediction model, vulnerable patches of the target sample are determined; The process of determining vulnerable patches of the target sample based on the fusion prediction model includes: Obtain the patch characteristics, risk factors, and radiomics characteristics of the target sample; The patch characteristics, risk factors, and radiomics characteristics of the target sample are input into the fusion prediction model to obtain evaluation parameters; The vulnerable plaques are determined based on the evaluation parameters; Patches whose evaluation parameters meet preset conditions are identified as vulnerable patches. The evaluation parameters include: AUC, sensitivity, specificity, accuracy, negative predictive value, and positive predictive value.
2. The method according to claim 1, characterized in that, Also includes: Acquire image data for at least two test samples, wherein the image data includes high-resolution MRI images before and after enhancement.
3. The method according to claim 2, characterized in that, Also includes: Based on the image data, patch information is determined. The plaque information includes: plaque diameter, minimum lumen area, intraplaque hemorrhage, minimum lumen diameter, stenosis rate, plaque burden, enhancement rate, and remodeling index.
4. The method according to claim 1, characterized in that, Also includes: A conventional model is constructed based on plaque characteristics and risk factors from at least two test samples, wherein the risk factors are used to characterize factors that are likely to lead to the formation of vulnerable plaques.
5. The method according to claim 1, characterized in that, The process of determining vulnerable patches of the target sample based on the fusion prediction model includes: Based on the fusion prediction model and the risk factors of the target sample, vulnerable plaques of the target sample are determined.
6. A device for detecting vulnerable plaques, characterized in that, The building blocks are used to construct radiomics models based on patch features from at least two experimental samples using plain scan plus enhanced 3D HRMR-VWI images; In ITK Snap software, the patch was delineated, and the 3D HRMR-VWI images before and after enhancement were imported into ITK SNAP to delineate and extract the ROI of the entire patch. An XGB model is constructed, which uses classification and regression trees as the basic classifier. The residual loss between the true value and the predicted value is reduced through iterative processing. The model is optimized along the gradient direction with the goal of reducing the residual loss, thereby continuously splitting features to grow the tree. Features were extracted from the ROIs obtained from the 3D HRMR-VWI before and after enhancement using PyRadiomics. The maximum correlation minimum redundancy radiomics feature selection was used, leaving 10 features for each sequence. The extreme gradient boosting method was used to establish a plain scan + enhanced 3D HRMR-VWI model. A unit was determined to determine the fusion prediction model based on the traditional model and the radiomics model. Traditional models are constructed based on the above plaque characteristics and risk factors; The indicators selected from clinical features and the 3D HRMR-VWI features before and after enhancement were used to construct the final fusion prediction model; The second determining unit is used to determine the vulnerable patches of the target sample based on the fusion prediction model; The process of determining vulnerable patches of the target sample based on the fusion prediction model includes: Obtain the patch characteristics, risk factors, and radiomics characteristics of the target sample; The patch characteristics, risk factors, and radiomics characteristics of the target sample are input into the fusion prediction model to obtain evaluation parameters; The vulnerable plaques are determined based on the evaluation parameters; Patches whose evaluation parameters meet preset conditions are identified as vulnerable patches. The evaluation parameters include: AUC, sensitivity, specificity, accuracy, negative predictive value, and positive predictive value.
7. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein, when the program is executed by a processor, it implements the steps of the vulnerable plaque detection method as described in any one of claims 1 to 5.
8. An electronic device, characterized in that, The electronic device includes at least one processor and at least one memory connected to the processor; wherein the processor is configured to invoke program instructions in the memory to execute the steps of the vulnerable plaque detection method as described in any one of claims 1 to 5.