Blood vessel wall thickness measurement method, device, equipment and storage medium

Abstract

The present disclosure provides a blood vessel wall thickness measurement method, device, equipment and storage medium, the method comprising: acquiring a first magnetic resonance image of cerebral blood vessels; determining the largest artery of the Willis circle from the first magnetic resonance image, and performing MPR image reconstruction along the centerline of the largest artery to obtain a first MPR image; determining the first skeleton point of the blood vessel in the first MPR image, and sampling with a preset voxel grid size with the first skeleton point as the center to obtain an image block corresponding to the first skeleton point; according to the image block corresponding to the first skeleton point, using a pre-constructed blood vessel wall thickness measurement model, calculating the blood vessel wall thickness value corresponding to the first skeleton point. The present disclosure can accurately determine the blood vessel wall thickness corresponding to the blood vessel skeleton point through conventional cerebral blood vessel magnetic resonance images, thereby improving the efficiency and accuracy of blood vessel wall thickness measurement.

Description

Technical Field

[0001] The present disclosure relates to the field of image processing technology, and in particular to a method, device, equipment and storage medium for measuring blood vessel wall thickness. Background Art

[0002] Intracranial arterial wall lesions are the root cause of cerebrovascular diseases. Among them, changes in intracranial vascular wall thickness are generally considered to provide important assistance for the diagnosis and differential diagnosis of cerebrovascular diseases (such as atherosclerosis). At present, vessel wall magnetic resonance imaging (VW-MRI) is the only non-invasive means of visualizing intracranial vascular walls. However, due to the low resolution of ordinary magnetic resonance images, it is impossible to accurately measure the thickness of the vascular wall in the very early stage. Although high-field magnetic resonance images have high resolution and can accurately measure the thickness of the vascular wall in the very early stage, their scanning time is long, which will cause discomfort to patients. In addition, the imaging equipment for obtaining high-field magnetic resonance images is expensive, and only a few hospitals have this technical equipment. Therefore, how to efficiently and accurately measure the thickness of the vascular wall is a technical problem that needs to be solved urgently by skilled technicians. Summary of the invention

[0003] In view of this, the present disclosure proposes a method, device, equipment and storage medium for measuring the thickness of a blood vessel wall, which can improve the efficiency and accuracy of measuring the thickness of a blood vessel wall.

[0004] According to a first aspect of the present disclosure, a method for measuring blood vessel wall thickness is provided, comprising:

[0005] acquiring a first magnetic resonance image of the cerebral vessels;

[0006] Determining the largest artery of the circle of Willis from the first magnetic resonance image, and performing MPR image reconstruction along the centerline of the largest artery to obtain a first MPR image;

[0007] Determine a first skeleton point of the blood vessel in the first MPR image, and take the first skeleton point as the center and perform sampling with a preset voxel grid size to obtain an image block corresponding to the first skeleton point;

[0008] According to the image block corresponding to the first skeleton point, a pre-constructed blood vessel wall thickness measurement model is used to calculate the blood vessel wall thickness value corresponding to the first skeleton point.

[0009] In a possible implementation manner, when determining a first skeleton point of a blood vessel in the first MPR image, the process includes:

[0010] Drawing the inner wall and the outer wall of the blood vessel in the first MPR image;

[0011] Based on the inner wall and the outer wall, a first skeleton point of the blood vessel in the first MPR image is determined.

[0012] In a possible implementation, the preset voxel grid size is 0.4 mm.

[0013] In one possible implementation, the vascular wall thickness measurement model is constructed based on a convolutional neural network architecture.

[0014] In a possible implementation, when constructing the blood vessel wall thickness measurement model based on a convolutional neural network architecture, it includes:

[0015] acquiring a second magnetic resonance image and a high-field strength magnetic resonance image of the cerebral blood vessels of the detection subject;

[0016] constructing sample data based on the second magnetic resonance image and the high-field strength magnetic resonance image, wherein the sample data includes image blocks corresponding to skeleton points and ground truth thickness;

[0017] The sample data is used to train the model parameters of the convolutional neural network architecture to obtain the blood vessel wall thickness measurement model.

[0018] In a possible implementation, when constructing sample data based on the second magnetic resonance image and the high-field strength magnetic resonance image, the method includes:

[0019] Acquire the high-field strength magnetic resonance image, determine the largest artery of the circle of Willis from the high-field strength magnetic resonance image, and perform MPR image reconstruction along the center line of the largest artery to obtain a third MPR image;

[0020] Determining a third skeleton point of the blood vessel in each of the third MPR images, and calculating a ground truth thickness corresponding to the third skeleton point;

[0021] Acquire the second magnetic resonance image, determine the largest artery of the circle of Willis from the second magnetic resonance image, and perform MPR image reconstruction along the centerline of the largest artery to obtain a second MPR image;

[0022] Acquire the third MPR image that matches the second MPR image, and annotate the second skeleton point of the blood vessel in the second MPR image and the ground measured thickness corresponding to the second skeleton point based on the third skeleton point of the blood vessel in the third MPR image that matches the second MPR image and the ground measured thickness corresponding to the third skeleton point;

[0023] Sample data is constructed based on second skeleton points of the blood vessels in the second MPR image and ground-measured thicknesses corresponding to the second skeleton points.

[0024] In a possible implementation, constructing sample data based on the second skeleton point of the blood vessel in the second MPR image and the ground measured thickness corresponding to the second skeleton point includes:

[0025] Taking the second skeleton point as the center and sampling with a preset voxel grid size, obtain an image block corresponding to the second skeleton point;

[0026] The ground measured thickness corresponding to the second skeleton point is used as the annotation data of the image block corresponding to the second skeleton point, and the image block annotated with the ground measured thickness is a sample data.

[0027] According to a second aspect of the present disclosure, a device for measuring blood vessel wall thickness is provided, comprising:

[0028] A magnetic resonance image acquisition module, used for acquiring a first magnetic resonance image of cerebral blood vessels;

[0029] An MPR image construction module, configured to determine the largest artery of the circle of Willis from the first magnetic resonance image, and perform MPR image reconstruction along the center line of the largest artery to obtain a first MPR image;

[0030] a sampling module, configured to determine a first skeleton point of a blood vessel in the first MPR image, and perform sampling with a preset voxel grid size with the first skeleton point as the center to obtain an image block corresponding to the first skeleton point;

[0031] The blood vessel thickness measurement module is used to calculate the blood vessel wall thickness value corresponding to the first skeleton point according to the image block corresponding to the first skeleton point using a pre-built blood vessel wall thickness measurement model.

[0032] According to a third aspect of the present disclosure, a device for measuring the thickness of a blood vessel wall is provided, comprising: a processor; and a memory for storing instructions executable by the processor; wherein the processor is configured to execute the method described in the first aspect of the present disclosure.

[0033] According to a fourth aspect of the present disclosure, a non-volatile computer-readable storage medium is provided, on which computer program instructions are stored, wherein the computer program instructions, when executed by a processor, implement the method described in the first aspect of the present disclosure.

[0034] In the present disclosure, a method for measuring the thickness of a blood vessel wall includes: acquiring a first magnetic resonance image of a cerebral blood vessel; determining the largest artery of the circle of Willis from the first magnetic resonance image, and performing MPR image reconstruction along the centerline of the largest artery to obtain a first MPR image; determining the first skeleton point of the blood vessel in the first MPR image, and sampling with a preset voxel grid size centered on the first skeleton point to obtain an image block corresponding to the first skeleton point; and calculating the blood vessel wall thickness value corresponding to the first skeleton point using a pre-constructed blood vessel wall thickness measurement model based on the image block corresponding to the first skeleton point. The present disclosure can accurately determine the blood vessel wall thickness corresponding to the blood vessel skeleton point through conventional cerebral blood vessel magnetic resonance images, thereby improving the efficiency and accuracy of blood vessel wall thickness measurement. It also enables hospitals without conditions for magnetic resonance enhancement examination to accurately measure the thickness of the blood vessel wall.

[0035] Further features and aspects of the present disclosure will become apparent from the following detailed description of exemplary embodiments with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate exemplary embodiments, features, and aspects of the disclosure and, together with the description, serve to explain the principles of the disclosure.

[0037] Figure 1 A flow chart showing a method for measuring blood vessel wall thickness according to an embodiment of the present disclosure;

[0038] Figure 2 A schematic diagram showing maximum artery tracking and MPR image reconstruction of the Willis circle according to an embodiment of the present disclosure is shown;

[0039] Figure 3 A structural diagram of a convolutional neural network architecture according to an embodiment of the present disclosure is shown;

[0040] Figure 4 A comparison diagram of a second MPR image and a corresponding third MPR image according to an embodiment of the present disclosure is shown;

[0041] Figure 5 A schematic block diagram showing a device for measuring the thickness of a blood vessel wall according to an embodiment of the present disclosure is shown;

[0042] Figure 6 A schematic block diagram of a device for measuring the thickness of a blood vessel wall according to an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0043] Various exemplary embodiments, features and aspects of the present disclosure will be described in detail below with reference to the accompanying drawings. The same reference numerals in the accompanying drawings represent elements with the same or similar functions. Although various aspects of the embodiments are shown in the accompanying drawings, the drawings are not necessarily drawn to scale unless otherwise specified.

[0044] The word “exemplary” is used exclusively herein to mean “serving as an example, example, or illustration.” Any embodiment described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other embodiments.

[0045] In addition, in order to better illustrate the present disclosure, numerous specific details are given in the following specific embodiments. It should be understood by those skilled in the art that the present disclosure can also be implemented without certain specific details. In some examples, methods, means, components and circuits well known to those skilled in the art are not described in detail in order to highlight the subject matter of the present disclosure.

[0046] <Method Example>

[0047] Figure 1 FIG. 5 is a flow chart showing a method for measuring the thickness of a blood vessel wall according to an embodiment of the present disclosure. Figure 1 As shown, the method includes steps S1100-S1400.

[0048] S1100, obtaining a first magnetic resonance image of the cerebral blood vessels, wherein the first magnetic resonance image is a magnetic resonance image obtained by a common magnetic resonance examination during the formal detection of the blood vessel wall thickness.

[0049] S1200, determining the largest artery of the circle of Willis from the first magnetic resonance image, and performing MPR image reconstruction along the center line of the largest artery of the circle of Willis to obtain a first MPR image. Specifically, after determining the largest artery of the circle of Willis from the first magnetic resonance image, tracking is performed along the center line of the largest artery using Euclidean distance transformation, as follows: Figure 2 As shown in A. During the tracking process, each time a center line point of a largest artery is tracked, an MPR image reconstruction is performed on the tracked center line point, thereby obtaining a first MPR image corresponding to the tracked center line point. When the tracking is completed, the first MPR image corresponding to each tracked center point will be obtained, that is, after the tracking is completed, multiple first MPR images will be obtained.

[0050] S1300, determine the first skeleton point of the blood vessel in the first MPR image, and take the first skeleton point as the center and perform sampling with a preset voxel grid size to obtain an image block corresponding to the first skeleton point. Specifically, for each first MPR image reconstructed in step S1200, perform the operations of steps S1310-S1320:

[0051] S1310, determining a first skeleton point of a blood vessel in a first MPR image, which may specifically include steps S1311-S132.

[0052] S1311, draw the inner wall and outer wall of the blood vessel in the first MPR image. Specifically, in the first MPR image: identify the inner wall boundary of the blood vessel, and then draw the inner wall of the blood vessel along the identified inner wall boundary. Figure 2 The inner wall of the blood vessel shown in B; the outer wall boundary of the blood vessel is identified, and then the outer wall boundary of the blood vessel is drawn along the identified outer wall boundary. Figure 2 The outer wall of the blood vessel shown in B. So far, the inner wall and the outer wall of the blood vessel in the first MPR image have been drawn.

[0053] Furthermore, in order to improve the accuracy of drawing the inner wall and the outer wall of the blood vessel, after the automatic drawing of the inner wall and the outer wall of the blood vessel is completed, calibration can be performed by manual modification.

[0054] S1312, based on the inner wall and outer wall of the drawn blood vessel, determine the first skeleton point of the blood vessel in the first MPR image. Specifically, based on the inner wall and outer wall of the drawn blood vessel wall, the first skeleton point in the blood vessel is calculated using 2D distance transformation. Figure 2 Multiple first skeleton points are shown in B.

[0055] In a possible implementation, the following steps may be used to calculate the first skeleton point in the blood vessel using 2D distance transformation according to the drawn inner wall and outer wall of the blood vessel:

[0056] First, the inner wall contour and outer wall contour of the blood vessel are extracted, wherein the inner wall contour represents the boundary of the lumen of the blood vessel, and the outer wall contour represents the boundary outside the blood vessel.

[0057] Secondly, a binary image is constructed based on the inner wall contour and outer wall contour of the blood vessel. Specifically, the area between the inner wall contour and the outer wall contour is marked as 1 (indicating the vascular lumen part); the area inside the inner wall contour and outside the outer wall contour is marked as 0 (indicating the non-vascular part).

[0058] Finally, the constructed binary image is subjected to 2D distance transformation to calculate the first skeleton point in the blood vessel. Specifically, the distance from each pixel to its nearest zero-value point (outer wall or background) can be obtained using the Euclidean distance transform. Then, the midline point between the inner and outer walls of the blood vessel is calculated, which is the first skeleton point.

[0059] S1320, taking the first skeleton point as the center, sampling with a preset voxel grid size to obtain an image block corresponding to the first skeleton point. The specific steps are as follows:

[0060] S1321, traverse the first skeleton in the first MPR image.

[0061] S1322, for the current first skeleton point traversed, take the current skeleton point as the center, and perform sampling with a preset voxel grid size to obtain an image block corresponding to the first skeleton point. Specifically, take the current skeleton point as the center, and perform isotropic sampling with a preset voxel grid size until an image block of a preset size is obtained. The value range of the preset voxel network size can be 0.2mm-1.0mm. Preferably, the voxel grid size can be set to 0.4mm. The value range of the image block size can be 9x9x9 voxels-51x51x51 voxels. Preferably, the size of the image block can be set to 19*19*19 voxels.

[0062] S1323, the traversal ends, and the image blocks corresponding to each first skeleton in the first MPR image are obtained. That is, the number of image blocks obtained by sampling is equal to the number of skeleton points in the first MPR image, and the number of image blocks obtained by sampling the first MPR image is equal to the number of skeleton points of the blood vessels in the first MPR image.

[0063] After the processing of steps S1310 - S1320 , for each first MPR image, multiple skeleton points of the blood vessels in the first MPR image are identified, and image blocks corresponding to the skeleton points are sampled.

[0064] S1400, according to the image block corresponding to the first skeleton point, using the pre-constructed vascular wall thickness measurement model, calculate the vascular wall thickness value corresponding to the first skeleton point. Specifically, for each first MPR image, the image block corresponding to each skeleton point is sequentially input into the pre-constructed vascular wall thickness measurement model, so that the vascular wall thickness measurement model predicts the vascular wall thickness value corresponding to the corresponding skeleton point according to the input image block, thereby completing the measurement of the vascular wall thickness corresponding to each skeleton point in the first MPR image through the vascular wall thickness measurement model.

[0065] It should be noted that before executing step S1400, step S1500 needs to be executed first to construct a blood vessel wall thickness measurement model for predicting the blood vessel wall thickness corresponding to the skeleton point according to the image block corresponding to the skeleton point.

[0066] In one possible implementation, the vascular wall thickness measurement model is constructed based on a convolutional neural network architecture. Figure 3 As shown in Figure 1, it includes two convolution paths, each of which consists of 5 convolution modules and a first fully connected layer (i.e. Figure 3 The first fully connected layers of the two convolution paths are connected to the second fully connected layer (i.e. Figure 3The output of the second fully connected layer is fed into the linearly activated output layer, which then outputs the final prediction result. The P value of the linearly activated output layer is 0.5 for the random dropout layer.

[0067] Each convolution module consists of a sequentially connected convolution layer Conv, a batch normalization layer BN, and an activation layer using a parameterized rectified linear unit pReLU. Figure 3 In the upper convolution path), the sizes of the feature maps input by the five convolution modules are 19*19*19, 13*13*13, 9*9*9, 7*7*7, 5*5*5, and the sizes of the convolution kernels are 7*7*7, 5*5*5, 3*3*3, 3*3*3, and 3*3*3. Figure 3 In the lower convolution path), the sizes of the feature maps input by the five convolution modules are 19*19*19, 15*15*15, 11*11*11, 7*7*7, and 5*5*5, and the sizes of the convolution kernels are 5*5*5, 5*5*5, 5*5*5, 5*5*5, and 3*3*3, respectively.

[0068] In constructing Figure 3 After the convolutional neural network architecture is shown, a vascular wall thickness measurement model can be constructed based on the convolutional neural network architecture. The steps of constructing the vascular wall thickness measurement model can include S1510-S1530.

[0069] S1510, obtaining a second magnetic resonance image and a high-field strength magnetic resonance image of the cerebral blood vessels of the test object. The second magnetic resonance image, i.e., the sample data for constructing the convolutional neural network architecture, is a magnetic resonance image obtained through an ordinary magnetic resonance examination. The high-field strength magnetic resonance image, i.e., the sample data for constructing the convolutional neural network architecture, is a magnetic resonance image obtained through an enhanced magnetic resonance examination. The clarity of the high-field strength magnetic resonance image is much higher than that of the first magnetic resonance image and the second magnetic resonance image.

[0070] Specifically, for the test object, a conventional MRI examination and an enhanced MRI examination are performed on the cerebral blood vessels of the test object, so that a second MRI image and a high-field strength MRI image of the same test object can be obtained. The number of test objects can be 1, 2, or more, which is not specifically limited here.

[0071] S1520, constructing sample data based on the second magnetic resonance image and the high-field strength magnetic resonance image of the same detection object, wherein the sample data includes image blocks corresponding to the skeleton points and ground truth thickness, wherein the ground truth thickness is the actual thickness value of the blood vessel wall corresponding to the skeleton points marked based on the image blocks corresponding to the skeleton points.

[0072] In a possible implementation, when constructing sample data based on the second magnetic resonance image and the high-field strength magnetic resonance image corresponding to the same detection, steps S1521 - S1525 may be included.

[0073] S1521, obtaining a high-field strength magnetic resonance image of the detection object, determining the largest artery of the Willis circle from the high-field strength magnetic resonance image, and performing MPR image reconstruction along the center line of the largest artery to obtain a third MPR image. The reconstruction process of the third MPR image refers to the reconstruction process of the first MPR image above, and will not be repeated here.

[0074] S1522, determining the third skeleton point of the blood vessel in each third MPR image, and calculating the ground truth thickness corresponding to the third skeleton point. Specifically, for each third MPR image, the following operations are performed:

[0075] First, the inner wall and the outer wall of the blood vessel are drawn in the third MPR image. The specific drawing method is shown in step S1311 and will not be described in detail here.

[0076] Second, based on the inner wall and the outer wall of the blood vessel in the third MPR image, the third skeleton points of the blood vessel in the third MPR image are determined, wherein for each third MPR image, the following skeleton points are determined: Figure 2 The method for determining the multiple third skeleton points shown in B refers to step S1312 and will not be described in detail here.

[0077] Third, traverse each third skeleton point of the blood vessel in the third MPR image, and for the current third skeleton point traversed, take the shortest distance between the inner wall and the outer wall of the current third skeleton point as the ground truth thickness corresponding to the current third skeleton point. When the traversal is completed, the ground truth thickness corresponding to each third skeleton point in the third MPR image is obtained.

[0078] S1523, obtaining a second magnetic resonance image of the same detection object, determining the largest artery of the Willis circle from the second magnetic resonance image, and performing MPR image reconstruction along the center line of the largest artery to obtain a second MPR image. The reconstruction process of the second MPR image refers to the reconstruction process of the first MPR image above, and will not be repeated here.

[0079] S1524, obtaining a third MPR image that matches the second MPR image, and marking the second skeleton points of the blood vessels in the second MPR image and the ground measured thickness corresponding to the second skeleton points based on the third skeleton points of the blood vessels in the third MPR image that matches the second MPR image and the ground measured thickness corresponding to the third skeleton points. Specifically, for the same detection object, multiple second MPR images and third MPR images are respectively acquired. For each second MPR image, a third MPR image that matches it is found. Specifically, for each second MPR image, a third MPR image with the same centerline point position is found as the third MPR image that matches it. For example, the 16 second MPR images reconstructed in the second magnetic resonance image can be as follows Figure 4 As shown in A, 16 third MPR images reconstructed from the high field strength magnetic resonance image and matching the second MPR image shown in 4A can be as follows Figure 4 As shown in Figure B, it can be seen that the resolution of the blood vessels in the third MPR image is much higher than that of the corresponding second MPR image. Therefore, the second skeleton points in the second MPR image and the ground measured thickness corresponding to the second skeleton points can be accurately marked based on the high-resolution third MPR image. Specifically, after finding the third MPR image that matches it, the third skeleton points in the third MPR image are traversed, and the pixel position and the corresponding ground measured thickness of the current third skeleton point are obtained for the current third skeleton point that has been traversed. In the second MPR image that matches the third MPR image, the position that is the same as the pixel position of the current third skeleton point is found as the position of the second skeleton point corresponding to the current third skeleton point, and the ground measured thickness corresponding to the current third skeleton point is used as the ground measured thickness corresponding to the second skeleton point corresponding to the current third skeleton point. After the traversal is completed, multiple second skeleton points can be marked in the second MPR image that matches the third MPR image, and the ground measured thickness corresponding to each second skeleton point can be determined.

[0080] S1525: construct sample data based on the second skeleton point of the blood vessel in the second MPR image and the ground measured thickness corresponding to the second skeleton point. Specifically, for each second skeleton point in the second MPR image, the following operations are performed:

[0081] First, taking the second skeleton point as the center, sampling is performed with a preset voxel grid size to obtain an image block corresponding to the second skeleton point. The specific image block acquisition process is shown in S1322 and will not be described in detail here.

[0082] Second, the ground measured thickness corresponding to the second skeleton point is used as the annotation data of the image block corresponding to the second skeleton point. The image block annotated with the ground measured thickness is a sample data, wherein the image block is the input image block of the convolutional neural network architecture, and the ground measured thickness annotated for the image block is the actual thickness of the blood vessel wall that needs to be predicted by the convolutional neural network architecture.

[0083] After completing the sampling of the image block corresponding to each second skeleton point, sample data corresponding to each second skeleton point will be obtained.

[0084] For each detected object, n second MPR images will be obtained, and for each MPR image, m sample data will be obtained, that is, for each detected object, n*m sample data will be obtained.

[0085] In order to improve the accuracy of model prediction, training samples of p-bit detection objects can be continuously collected to obtain p*n*m sample data. Then, data enhancement and rotation processing are performed on the p*n*m sample data to expand the sample data until the total number of samples reaches the preset training number. Among them, the preset training number can be determined according to specific needs. For example, the preset training number can be set to 3,000,000, that is, 300,000 image blocks with the measured thickness of the blood vessels on the ground are constructed as sample data of the blood vessel wall thickness measurement model.

[0086] S1530, using sample data to train the model parameters of the convolutional neural network architecture to obtain a vascular wall thickness measurement model. Specifically, after obtaining the sample data, the sample data is randomly divided into a training set and a validation set, wherein the number ratio of the sample data in the training set and the validation set is 2:1. Next, the sample data in the training set is first used to train the model parameters of the convolutional neural network architecture in the above text to obtain an initial vascular wall thickness measurement model. After obtaining the initial vascular wall thickness measurement model, the sample data in the validation set is used to verify the model until the model error does not decrease within 80 periods, and the model obtained at this time is used as the final trained vascular wall thickness measurement model.

[0087] After obtaining the vascular wall thickness measurement model, the image block corresponding to the first skeleton point is input into the vascular wall thickness measurement model, and the vascular wall thickness measurement model will extract features from the image block through the 5 convolution modules in the first convolution path to obtain the first feature map with edge highlight enhancement; extract features from the image block through the 5 convolution modules in the second convolution path to obtain the second feature map with intensity flipping and wall enhancement; the first feature map and the second feature map are fed into the linearly activated output layer after being processed by the first fully connected layer and the second fully connected layer, and the linearly activated output layer outputs the predicted vascular wall thickness value for the image block. Among them, the first fully connected layer, the second fully connected layer and the linearly activated output layer can be neural networks preset by the user. It is worth mentioning that the linearly activated output layer is constructed based on the preset activation function for the regression task.

[0088] The present disclosure provides a method for measuring the thickness of a blood vessel wall, including: acquiring a first magnetic resonance image of a cerebral blood vessel; determining the largest artery of the circle of Willis from the first magnetic resonance image, and performing MPR image reconstruction along the centerline of the largest artery to obtain a first MPR image; determining the first skeleton point of the blood vessel in the first MPR image, and sampling with a preset voxel grid size with the first skeleton point as the center to obtain an image block corresponding to the first skeleton point; and calculating the blood vessel wall thickness value corresponding to the first skeleton point using a pre-constructed blood vessel wall thickness measurement model based on the image block corresponding to the first skeleton point. The present disclosure can accurately determine the blood vessel wall thickness corresponding to the blood vessel skeleton point through conventional cerebral blood vessel magnetic resonance images, thereby improving the efficiency and accuracy of blood vessel wall thickness measurement. It also enables hospitals without magnetic resonance enhancement examination conditions to accurately measure the thickness of the blood vessel wall.

[0089] <Device Example>

[0090] Figure 5 FIG. 1 is a schematic block diagram of a device for measuring the thickness of a blood vessel wall according to an embodiment of the present disclosure. Figure 5 As shown, the device 100 includes:

[0091] A magnetic resonance image acquisition module 110, configured to acquire a first magnetic resonance image of cerebral blood vessels;

[0092] An MPR image construction module 120 is used to determine the largest artery of the circle of Willis from the first magnetic resonance image, and perform MPR image reconstruction along the center line of the largest artery to obtain a first MPR image;

[0093] A sampling module 130 is used to determine a first skeleton point of a blood vessel in the first MPR image, and to perform sampling with a preset voxel grid size with the first skeleton point as the center to obtain an image block corresponding to the first skeleton point;

[0094] The blood vessel thickness measurement module 140 is used to calculate the blood vessel wall thickness value corresponding to the first skeleton point according to the image block corresponding to the first skeleton point by using a pre-built blood vessel wall thickness measurement model.

[0095] <Equipment Embodiment>

[0096] Figure 6 FIG. 1 is a schematic block diagram of a blood vessel wall thickness measuring device according to an embodiment of the present disclosure. Figure 6 As shown, the blood vessel wall thickness measuring device 200 includes: a processor 210 and a memory 220 for storing executable instructions of the processor 210. The processor 210 is configured to implement any of the above-mentioned blood vessel wall thickness measuring methods when executing the executable instructions.

[0097] Here, it should be noted that the number of processors 210 can be one or more. At the same time, in the vascular wall thickness measurement device 200 of the embodiment of the present disclosure, an input device 230 and an output device 240 may also be included. The processor 210, the memory 220, the input device 230 and the output device 240 may be connected via a bus or in other ways, which are not specifically limited here.

[0098] The memory 220 is a computer-readable storage medium that can be used to store software programs, computer executable programs, and various modules, such as the program or module corresponding to the vascular wall thickness measurement method of the embodiment of the present disclosure. The processor 210 executes various functional applications and data processing of the vascular wall thickness measurement device 200 by running the software programs or modules stored in the memory 220.

[0099] The input device 230 may be used to receive input numbers or signals. The signals may be key signals related to user settings and function control of the device / terminal / server. The output device 240 may include display devices such as display screens.

[0100] <Storage Medium Embodiment>

[0101] According to a fourth aspect of the present disclosure, a non-volatile computer-readable storage medium is also provided, on which computer program instructions are stored. When the computer program instructions are executed by the processor 210, any of the above-mentioned methods for measuring the thickness of the blood vessel wall is implemented.

[0102] The embodiments of the present disclosure have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or technical improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. A method for measuring blood vessel wall thickness, characterized in that: include: acquiring a first magnetic resonance image of the cerebral vessels; Determining the largest artery of the circle of Willis from the first magnetic resonance image, and performing MPR image reconstruction along the centerline of the largest artery to obtain a first MPR image; Determine a first skeleton point of the blood vessel in the first MPR image, and take the first skeleton point as the center and perform sampling with a preset voxel grid size to obtain an image block corresponding to the first skeleton point; Calculate the blood vessel wall thickness value corresponding to the first skeleton point according to the image block corresponding to the first skeleton point using a pre-constructed blood vessel wall thickness measurement model; The blood vessel wall thickness measurement model is constructed based on a convolutional neural network architecture; When constructing the vascular wall thickness measurement model based on the convolutional neural network architecture, it includes: acquiring a second magnetic resonance image and a high-field strength magnetic resonance image of the cerebral blood vessels of the detection subject; constructing sample data based on the second magnetic resonance image and the high-field strength magnetic resonance image, wherein the sample data includes image blocks corresponding to skeleton points and ground truth thickness; Using the sample data to train the model parameters of the convolutional neural network architecture to obtain the blood vessel wall thickness measurement model; When constructing sample data based on the second magnetic resonance image and the high-field strength magnetic resonance image, it includes: Acquire the high-field strength magnetic resonance image, determine the largest artery of the circle of Willis from the high-field strength magnetic resonance image, and perform MPR image reconstruction along the center line of the largest artery to obtain a third MPR image; Determining a third skeleton point of the blood vessel in each of the third MPR images, and calculating a ground truth thickness corresponding to the third skeleton point; Acquire the second magnetic resonance image, determine the largest artery of the circle of Willis from the second magnetic resonance image, and perform MPR image reconstruction along the centerline of the largest artery to obtain a second MPR image; Acquire the third MPR image that matches the second MPR image, and annotate the second skeleton point of the blood vessel in the second MPR image and the ground measured thickness corresponding to the second skeleton point based on the third skeleton point of the blood vessel in the third MPR image that matches the second MPR image and the ground measured thickness corresponding to the third skeleton point; Sample data is constructed based on second skeleton points of the blood vessels in the second MPR image and ground-measured thicknesses corresponding to the second skeleton points.

2. The method according to claim 1, characterized in that When determining a first skeleton point of a blood vessel in the first MPR image, the method includes: Drawing the inner wall and the outer wall of the blood vessel in the first MPR image; Based on the inner wall and the outer wall, a first skeleton point of the blood vessel in the first MPR image is determined.

3. The method according to claim 1, characterized in that The preset voxel grid size is 0.4 mm.

4. The method according to claim 1, characterized in that When constructing sample data based on the second skeleton point of the blood vessel in the second MPR image and the ground measured thickness corresponding to the second skeleton point, the method includes: Taking the second skeleton point as the center and sampling with a preset voxel grid size, obtain an image block corresponding to the second skeleton point; The ground measured thickness corresponding to the second skeleton point is used as the annotation data of the image block corresponding to the second skeleton point, and the image block annotated with the ground measured thickness is a sample data.

5. A blood vessel wall thickness measuring device, characterized in that: include: A magnetic resonance image acquisition module, used for acquiring a first magnetic resonance image of cerebral blood vessels; An MPR image construction module, configured to determine the largest artery of the circle of Willis from the first magnetic resonance image, and perform MPR image reconstruction along the center line of the largest artery to obtain a first MPR image; a sampling module, configured to determine a first skeleton point of a blood vessel in the first MPR image, and to perform sampling with a preset voxel grid size with the first skeleton point as the center to obtain an image block corresponding to the first skeleton point; a blood vessel thickness measurement module, configured to calculate the blood vessel wall thickness value corresponding to the first skeleton point according to the image block corresponding to the first skeleton point by using a pre-built blood vessel wall thickness measurement model; The blood vessel wall thickness measurement model is constructed based on a convolutional neural network architecture; When constructing the vascular wall thickness measurement model based on the convolutional neural network architecture, it includes: acquiring a second magnetic resonance image and a high-field strength magnetic resonance image of the cerebral blood vessels of the detection subject; constructing sample data based on the second magnetic resonance image and the high-field strength magnetic resonance image, wherein the sample data includes image blocks corresponding to skeleton points and ground truth thickness; Using the sample data to train the model parameters of the convolutional neural network architecture to obtain the blood vessel wall thickness measurement model; When constructing sample data based on the second magnetic resonance image and the high-field strength magnetic resonance image, it includes: Acquire the high-field strength magnetic resonance image, determine the largest artery of the circle of Willis from the high-field strength magnetic resonance image, and perform MPR image reconstruction along the center line of the largest artery to obtain a third MPR image; Determining a third skeleton point of the blood vessel in each of the third MPR images, and calculating a ground truth thickness corresponding to the third skeleton point; Acquire the second magnetic resonance image, determine the largest artery of the circle of Willis from the second magnetic resonance image, and perform MPR image reconstruction along the centerline of the largest artery to obtain a second MPR image; Acquire the third MPR image that matches the second MPR image, and annotate the second skeleton point of the blood vessel in the second MPR image and the ground measured thickness corresponding to the second skeleton point based on the third skeleton point of the blood vessel in the third MPR image that matches the second MPR image and the ground measured thickness corresponding to the third skeleton point; Sample data is constructed based on second skeleton points of the blood vessels in the second MPR image and ground-measured thicknesses corresponding to the second skeleton points.

6. A blood vessel wall thickness measuring device, characterized in that: include: processor; a memory for storing processor-executable instructions; Wherein, the processor is configured to implement the method described in any one of claims 1 to 4 when executing the executable instructions.

7. A non-volatile computer-readable storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the method according to any one of claims 1 to 4 is implemented.

Images