Plaque identification method, device, computer equipment and storage medium

Abstract

The present application relates to a plaque recognition method, device, computer equipment and storage medium, and relates to the field of medical image analysis technology. The plaque recognition method obtains a set of cross-sectional images corresponding to a target cross-sectional image, performs histogram equalization processing on each cross-sectional image in the cross-sectional image set, and obtains an equalized image corresponding to each cross-sectional image; multiple cross-sectional images and the equalized images corresponding to each cross-sectional image are input into a trained plaque segmentation model to obtain a plaque recognition result of the target cross-sectional image. In this method, when identifying plaques in a target cross-sectional image, analysis and recognition are performed based on multiple cross-sectional images included in the cross-sectional image set corresponding to the target cross-sectional image. Therefore, contextual information corresponding to the target cross-sectional image can be obtained, and more information is obtained, so the plaque recognition result is more accurate.

Description

Technical Field

[0001] The present application relates to the technical field of medical image analysis, and in particular to a plaque recognition method, apparatus, computer equipment, and storage medium. Background Art

[0002] Carotid artery plaque is one of the important causes of cerebral stroke, so the detection of carotid artery plaque is particularly important. In the process of carotid artery plaque detection, it is necessary to perform carotid artery plaque segmentation on the carotid artery image.

[0003] Existing carotid artery plaque segmentation solutions generally include: performing image recognition processing on a certain carotid artery image to determine whether there is a plaque in the carotid artery image and the location of the plaque.

[0004] However, in the above method, during the plaque segmentation process, only information in the carotid artery plaque image can be obtained, and the obtained information is relatively small, resulting in inaccurate segmentation results. Summary of the Invention

[0005] Based on this, it is necessary to provide a plaque recognition method, apparatus, computer equipment and storage medium that can improve the accuracy of plaque recognition in order to address the above technical problems.

[0006] A plaque identification method, the method comprising:

[0007] Acquire a cross-sectional image set corresponding to a target cross-sectional image, where the cross-sectional image set includes multiple cross-sectional images, and the target cross-sectional image is one of the multiple cross-sectional images;

[0008] Performing histogram equalization processing on each cross-sectional image in the cross-sectional image set to obtain an equalized image corresponding to each cross-sectional image;

[0009] Multiple cross-sectional images and the equalized images corresponding to each cross-sectional image are input into the trained plaque segmentation model to obtain the plaque recognition result of the target cross-sectional image.

[0010] In one embodiment, obtaining a set of cross-sectional images corresponding to a target cross-sectional image includes:

[0011] Obtain a target cross-sectional image from a centerline point on the centerline of the blood vessel and a cross-sectional image perpendicular to the centerline of the blood vessel, including the centerline point in the three-dimensional image;

[0012] Determine multiple candidate center points on both sides of the center point on the blood vessel center line, and obtain a cross-sectional image corresponding to each candidate center point;

[0013] A set of cross-sectional images corresponding to the target cross-sectional image is obtained according to the target cross-sectional image and the cross-sectional images corresponding to each candidate center point.

[0014] In one embodiment, a plurality of candidate center points are determined on both sides of a center point on the blood vessel center line, including:

[0015] Taking the centerline point as the starting point, multiple candidate center points are determined on both sides of the centerline point on the centerline of the blood vessel based on a preset step size.

[0016] In one embodiment, the method further comprises:

[0017] Identify and process the blood vessel wall in the target cross-sectional image to obtain a vessel wall annotation map;

[0018] Multiple cross-sectional images and the equalized images corresponding to each cross-sectional image are input into the trained plaque segmentation model to obtain the plaque recognition results of the target cross-sectional image, including:

[0019] The pipe wall annotation map, multiple cross-sectional images and the equalized images corresponding to each cross-sectional image are input into the trained plaque segmentation model to obtain the plaque recognition result of the target cross-sectional image.

[0020] In one embodiment, the plaque segmentation model includes a first plaque segmentation model, which inputs a vessel wall annotation image, multiple cross-sectional images, and equalized images corresponding to each cross-sectional image into the trained plaque segmentation model to obtain a plaque recognition result of the target cross-sectional image, including:

[0021] After normalizing the multiple cross-sectional images and the equalized images corresponding to the cross-sectional images, the pipe wall annotation image, the multiple cross-sectional images, and the equalized images corresponding to the cross-sectional images are merged to obtain first multi-channel input data;

[0022] The first multi-channel input data is input into a first plaque segmentation model to obtain a first plaque recognition result of the target cross-sectional image.

[0023] In one embodiment, the plaque segmentation model further includes a plaque recognition model and a second plaque segmentation model. The pipe wall annotation map, multiple cross-sectional images, and the equalized images corresponding to each cross-sectional image are input into the trained plaque segmentation model to obtain a plaque recognition result of the target cross-sectional image, including:

[0024] After normalizing the multiple cross-sectional images, the pipe wall annotation image and the multiple cross-sectional images are merged to obtain second multi-channel input data;

[0025] inputting the second multi-channel input data into the plaque recognition model to obtain plaque types of plaques in the target cross-sectional image;

[0026] If the plaque type of the plaque in the target cross-sectional image is a preset type, inputting the second multi-channel input data into the second plaque segmentation model to obtain a second plaque recognition result of the target cross-sectional image;

[0027] A plaque recognition result of the target cross-sectional image is obtained according to the first plaque recognition result and the second plaque recognition result.

[0028] In one embodiment, obtaining a plaque recognition result of the target cross-sectional image according to the first plaque recognition result and the second plaque recognition result includes:

[0029] The first plaque recognition result and the second plaque recognition result are taken as a union to obtain the plaque recognition result of the target cross-sectional image.

[0030] A plaque identification device, comprising:

[0031] an acquisition module, configured to acquire a cross-sectional image set corresponding to a target cross-sectional image, wherein the cross-sectional image set includes a plurality of cross-sectional images, and the target cross-sectional image is one of the plurality of cross-sectional images;

[0032] a processing module, configured to perform histogram equalization processing on each cross-sectional image in the cross-sectional image set to obtain an equalized image corresponding to each cross-sectional image;

[0033] The segmentation module is used to input multiple cross-sectional images and the equalized images corresponding to each cross-sectional image into the trained plaque segmentation model to obtain the plaque recognition result of the target cross-sectional image.

[0034] A computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are performed:

[0035] Acquire a cross-sectional image set corresponding to a target cross-sectional image, where the cross-sectional image set includes multiple cross-sectional images, and the target cross-sectional image is one of the multiple cross-sectional images;

[0036] Performing histogram equalization processing on each cross-sectional image in the cross-sectional image set to obtain an equalized image corresponding to each cross-sectional image;

[0037] Multiple cross-sectional images and the equalized images corresponding to each cross-sectional image are input into the trained plaque segmentation model to obtain the plaque recognition result of the target cross-sectional image.

[0038] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the following steps:

[0039] Acquire a cross-sectional image set corresponding to a target cross-sectional image, where the cross-sectional image set includes multiple cross-sectional images, and the target cross-sectional image is one of the multiple cross-sectional images;

[0040] Performing histogram equalization processing on each cross-sectional image in the cross-sectional image set to obtain an equalized image corresponding to each cross-sectional image;

[0041] Multiple cross-sectional images and the equalized images corresponding to each cross-sectional image are input into the trained plaque segmentation model to obtain the plaque recognition result of the target cross-sectional image.

[0042] The aforementioned plaque recognition method, apparatus, computer device, and storage medium can improve plaque recognition accuracy. This plaque recognition method obtains a set of cross-sectional images corresponding to a target cross-sectional image, performs histogram equalization on each cross-sectional image in the set, and obtains an equalized image corresponding to each cross-sectional image. Multiple cross-sectional images and the equalized images corresponding to each cross-sectional image are then input into a trained plaque segmentation model to obtain a plaque recognition result for the target cross-sectional image. This method identifies plaques in a target cross-sectional image based on analysis and recognition of the multiple cross-sectional images included in the set of cross-sectional images corresponding to the target cross-sectional image. This method therefore obtains contextual information corresponding to the target cross-sectional image, providing more information and resulting in more accurate plaque recognition results. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 1 is a flow chart of a plaque identification method according to an embodiment;

[0044] Figure 2 is a flow chart of a plaque identification method according to another embodiment;

[0045] Figure 3 is a structural block diagram of a plaque identification device in one embodiment;

[0046] Figure 4 FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solutions and advantages of this application more clearly understood, this application is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely for the purpose of explaining this application and are not intended to limit this application. In addition, the technical features involved in the different embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0048] In one embodiment, Figure 1As shown, a plaque recognition method is provided. This embodiment uses the method applied to a terminal as an example for illustration. It is understandable that the method can also be applied to a server, or to a system including a terminal and a server, and implemented through interaction between the terminal and the server. The terminal is a medical ultrasound device, a medical scanning device, or other medical device. In this embodiment, the method includes the following steps:

[0049] Step 101: Acquire a set of cross-sectional images corresponding to a target cross-sectional image.

[0050] The cross-sectional image set includes multiple cross-sectional images, and the target cross-sectional image is one of the multiple cross-sectional images.

[0051] In the embodiment of the present application, a three-dimensional stereoscopic image may be acquired, and the three-dimensional stereoscopic image may be, for example, a three-dimensional carotid artery image, a three-dimensional coronary artery image, etc. The following description will be made using a three-dimensional carotid artery image as an example.

[0052] In an optional implementation, for the acquired three-dimensional carotid artery image, multiple cross-sectional images are extracted along the centerline of the blood vessel and sorted in order according to their positions in the three-dimensional carotid artery image. Then, partial cross-sectional images arranged in order are extracted from the multiple cross-sectional images to form a cross-sectional image set, and each cross-sectional image set contains a target cross-sectional image, wherein the target cross-sectional image can be any image in the cross-sectional image set, or it can be a cross-sectional image in the middle position in the cross-sectional image set.

[0053] In another optional implementation, for the acquired three-dimensional carotid artery image, multiple centerline points are selected along the centerline of the blood vessel. For each centerline point, a cross-sectional image perpendicular to the centerline of the blood vessel, including the centerline point, is intercepted from the three-dimensional carotid artery image to obtain a target cross-sectional image. Then, with the centerline point as the center, multiple candidate center points are determined along the front-to-back direction of the blood vessel centerline, and the cross-sectional image corresponding to each candidate center point is intercepted from the three-dimensional carotid artery image. Based on the target cross-sectional image and the cross-sectional images corresponding to each candidate center point, a set of cross-sectional images corresponding to the target cross-sectional image is obtained.

[0054] Optionally, in an embodiment of the present application, the centerline point can be used as a starting point, and multiple candidate center points can be obtained on both sides of the centerline point on the centerline of the blood vessel with preset step sizes.

[0055] For example, at the centerline point P, n candidate center points are obtained along the front and rear directions of the blood vessel centerline with a step size of λ, for a total of (2n+1) points. With these (2n+1) points as the center and the XY axis of the centerline point P as the direction, (2n+1) cross-sectional images are reconstructed to form a cross-sectional image set corresponding to the target cross-sectional image corresponding to the centerline point P.

[0056] Step 102 : Perform histogram equalization processing on each cross-sectional image in the cross-sectional image set to obtain a balanced image corresponding to each cross-sectional image.

[0057] In the embodiment of the present application, for each cross-sectional image in the cross-sectional image set, a histogram equalization algorithm is used to map the grayscale value of the cross-sectional image to obtain an equalized image corresponding to each cross-sectional image.

[0058] Step 103 : Input the multiple cross-sectional images and the equalized images corresponding to the cross-sectional images into the trained plaque segmentation model to obtain the plaque recognition result of the target cross-sectional image.

[0059] In an embodiment of the present application, a plurality of cross-sectional images and the equalized images corresponding to the cross-sectional images may be sequentially input into a trained plaque segmentation model.

[0060] Optionally, each cross-sectional image and its corresponding equalized image may be put together, and then arranged in the order of the multiple cross-sectional images, and then sequentially input into the trained plaque segmentation model.

[0061] Optionally, in an embodiment of the present application, for a target training cross-sectional image, a single-channel model can be used to perform plaque segmentation on the target training cross-sectional image, and the segmented image is used as a gold standard G. Then, a multi-channel segmentation model is obtained by training a set of training cross-sectional images corresponding to the target training cross-sectional image and the gold standard G based on a segmentation network and a segmentation loss. The trained plaque segmentation model is a multi-channel segmentation model.

[0062] Optionally, in an embodiment of the present application, before inputting the multiple cross-sectional images and the equalized images corresponding to each cross-sectional image into the plaque segmentation model, each cross-sectional image and the equalized image corresponding to each cross-sectional image may be normalized separately, and then the multiple normalized images may be input into the plaque segmentation model.

[0063] In the embodiment of the present application, since the equalized image can strengthen the plaque boundary, and the multiple cross-sectional images increase the amount of plaque information obtained by the plaque segmentation model, the recognition accuracy of the plaque segmentation model is higher.

[0064] In an optional implementation, the blood vessel wall in the target cross-sectional image is identified and processed to obtain a wall annotation map; the wall annotation map, multiple cross-sectional images, and the equalized images corresponding to each cross-sectional image are input into a trained plaque segmentation model to obtain the plaque recognition result of the target cross-sectional image.

[0065] The vessel wall annotation image is an image created by annotating the contours of the vessel wall in the target cross-sectional image. This annotation process uses lines to depict the vessel wall contours in the target cross-sectional image, allowing the plaque segmentation model to accurately identify the vessel wall's location. Plaques are always located on the inner side of the vessel wall, allowing for rapid identification of plaque locations, reducing unnecessary computation and improving efficiency.

[0066] In an embodiment of the present application, a Gaussian model can be performed on the blood vessel wall in the target cross-sectional image to obtain a wall annotation map, and then the wall annotation map and multiple cross-sectional images and the equalized images corresponding to the multiple cross-sectional images are merged into a first multi-channel input data, and then the first multi-channel input data is input into a plaque segmentation model to obtain a plaque segmentation result.

[0067] In one optional implementation, the plaque segmentation model includes a first plaque segmentation model, which is used to perform preliminary plaque identification. In an embodiment of the present application, after normalizing the pipe wall annotation map, multiple cross-sectional images, and equalized images corresponding to each cross-sectional image, the pipe wall annotation map, multiple cross-sectional images, and equalized images corresponding to each cross-sectional image are merged to obtain first multi-channel input data. The first multi-channel input data is then input into the first plaque segmentation model to obtain a plaque identification result output by the first plaque segmentation model.

[0068] The process of merging the pipe wall annotation map, the multiple cross-sectional images, and the equalized images corresponding to each cross-sectional image may be, for example, placing each cross-sectional image and its corresponding equalized image together, and then sorting the multiple cross-sectional images in the order of precedence, wherein the pipe wall annotation map is placed together with the target cross-sectional image and the equalized image corresponding to the target cross-sectional image. Alternatively, the pipe wall annotation map, the multiple cross-sectional images, and the multiple equalized images may be arranged together in sequence.

[0069] The present invention utilizes 1. Multi-layer reconstruction: Reconstructing multiple layers of vascular cross-sectional images forward and backward along the vessel at the centerline point as multi-channel input, while maintaining the lightweight nature of the 2D segmentation network while introducing 3D information, improving the network's field of view and the robustness of the segmentation results. 2. Histogram equalization: Improving the contrast of plaques, thereby improving segmentation accuracy. 3. Wall restriction: Introducing wall restriction allows the network to focus on the vessel wall region, while introducing a manually defined attention mechanism to ensure that the segmentation results are within the vessel wall.

[0070] Since different types of plaques have different corresponding signal characteristics, this feature can be used to specifically identify some specific plaques to improve the accuracy of plaque identification. In an optional implementation, such as Figure 2 As shown, the plaque segmentation model also includes a plaque recognition model and a second plaque segmentation model; wherein the plaque recognition model is used to identify the type of plaque, and the second plaque segmentation model is used to mark the plaque.

[0071] Step 201 : After normalizing the multiple cross-sectional images, the pipe wall annotation image and the multiple cross-sectional images are merged to obtain second multi-channel input data.

[0072] In the embodiment of the present application, the second channel input data does not include equalized images of each cross-sectional image. This is because equalized images sacrifice image information to enhance image contrast, which is beneficial for locating plaques but not for identifying plaque types. Therefore, when performing plaque type identification, the equalized images of each cross-sectional image are discarded.

[0073] Optionally, in the second multi-channel input data, the vessel wall annotation image and the multiple cross-sectional images may be arranged such that the vessel wall annotation image is placed first, and the multiple cross-sectional images are arranged sequentially below the vessel wall annotation image. Alternatively, the vessel wall annotation image may be placed after the multiple cross-sectional images. Alternatively, the vessel wall annotation image may be placed adjacent to the target cross-sectional image.

[0074] Step 202 : Input the second multi-channel input data into a plaque recognition model to obtain the plaque type of the plaque in the target cross-sectional image.

[0075] In the embodiment of the present application, the plaque recognition module is a multi-channel recognition model, which can process multiple images included in the second multi-channel input data to obtain the plaque type of the plaque in the target cross-sectional image.

[0076] Optionally, in an embodiment of the present application, the plaque recognition model may be a trained recognition model for specific plaques, for example, the plaque recognition model may be a calcified plaque recognition model, which is used to specifically recognize calcified plaques in a target cross-sectional image.

[0077] Step 203 : If the plaque type of the plaque in the target cross-sectional image is a preset type, the second multi-channel input data is input into a second plaque segmentation model to obtain a second plaque recognition result of the target cross-sectional image.

[0078] If the plaque in the target cross-sectional image is a preset type of plaque, such as a calcified plaque, the second multi-channel data is input into a second plaque segmentation model, wherein the second plaque segmentation model is used to specifically segment the preset type of plaque.

[0079] If the plaque in the target cross-sectional image is not a plaque of the preset type, the second multi-channel input data is discarded, and the first plaque recognition result is used as the plaque recognition result of the target cross-sectional image.

[0080] Step 204 : Obtain a plaque recognition result of the target cross-sectional image according to the first plaque recognition result and the second plaque recognition result.

[0081] In the embodiment of the present application, the first plaque recognition result is generally the recognition of ordinary plaques, and the second plaque recognition result is the result obtained by specifically identifying the preset type of plaque when it is determined that the plaque in the target cross-sectional image is a preset type of plaque, so the recognition result is more accurate.

[0082] Optionally, when the plaque in the target cross-sectional image is a preset type of plaque, the second plaque recognition result is used as the plaque recognition result of the target cross-sectional image.

[0083] Optionally, in an embodiment of the present application, the first plaque recognition result and the second plaque recognition result are taken as a union to obtain a plaque recognition result of the target cross-sectional image.

[0084] In the embodiments of the present application, a plaque recognition model is used to identify a preset type of plaque. If a plaque in a target cross-sectional image is of the preset type, the plaque in the target cross-sectional image is segmented using the second multi-channel input data and the second plaque segmentation model, thereby improving the accuracy of plaque identification. For example, if the preset type of plaque is a calcified plaque, and the low signal characteristics of calcification on multiple sequences are not conducive to segmentation, the calcification segmentation model is used to segment the plaque when a calcified plaque is identified, thereby specifically segmenting the calcified plaque and improving the overall accuracy of plaque segmentation.

[0085] It should be understood that although Figure 1-2 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. In addition, Figure 1-2At least part of the steps may include multiple steps or multiple stages. These steps or stages are not necessarily performed at the same time, but can be performed at different times. The order of execution of these steps or stages is not necessarily one by one, but can be performed in turn or alternately with other steps or at least part of the steps or stages in other steps.

[0086] In one embodiment, Figure 3 As shown, a plaque identification device 300 is provided, comprising: an acquisition module 301, a processing module 302 and a segmentation module 303, wherein:

[0087] An acquisition module 301 is configured to acquire a cross-sectional image set corresponding to a target cross-sectional image, wherein the cross-sectional image set includes multiple cross-sectional images, and the target cross-sectional image is one of the multiple cross-sectional images;

[0088] The processing module 302 is configured to perform histogram equalization processing on each cross-sectional image in the cross-sectional image set to obtain an equalized image corresponding to each cross-sectional image;

[0089] The segmentation module 303 is used to input the multiple cross-sectional images and the equalized images corresponding to the cross-sectional images into the trained plaque segmentation model to obtain the plaque recognition result of the target cross-sectional image.

[0090] In one embodiment, the acquisition module 301 is specifically configured to:

[0091] Obtain a target cross-sectional image from a centerline point on the centerline of the blood vessel and a cross-sectional image perpendicular to the centerline of the blood vessel, including the centerline point in the three-dimensional image;

[0092] Determine multiple candidate center points on both sides of the center point on the blood vessel center line, and obtain a cross-sectional image corresponding to each candidate center point;

[0093] A set of cross-sectional images corresponding to the target cross-sectional image is obtained according to the target cross-sectional image and the cross-sectional images corresponding to each candidate center point.

[0094] In one embodiment, the acquisition module 301 is specifically configured to:

[0095] Taking the centerline point as the starting point, multiple candidate center points are determined on both sides of the centerline point on the centerline of the blood vessel based on a preset step size.

[0096] In one embodiment, the segmentation module 303 is specifically configured to:

[0097] Identify and process the blood vessel wall in the target cross-sectional image to obtain a vessel wall annotation map;

[0098] The pipe wall annotation map, multiple cross-sectional images and the equalized images corresponding to each cross-sectional image are input into the trained plaque segmentation model to obtain the plaque recognition result of the target cross-sectional image.

[0099] In one embodiment, the plaque segmentation model includes a first plaque segmentation model, and the segmentation module 303 is specifically configured to:

[0100] After normalizing the multiple cross-sectional images and the equalized images corresponding to the cross-sectional images, the pipe wall annotation image, the multiple cross-sectional images, and the equalized images corresponding to the cross-sectional images are merged to obtain first multi-channel input data;

[0101] The first multi-channel input data is input into a first plaque segmentation model to obtain a first plaque recognition result of the target cross-sectional image.

[0102] In one embodiment, the plaque segmentation model further includes a plaque recognition model and a second plaque segmentation model, and the segmentation module 303 is specifically configured to:

[0103] After normalizing the multiple cross-sectional images, the pipe wall annotation image and the multiple cross-sectional images are merged to obtain second multi-channel input data;

[0104] inputting the second multi-channel input data into the plaque recognition model to obtain plaque types of plaques in the target cross-sectional image;

[0105] If the plaque type of the plaque in the target cross-sectional image is a preset type, inputting the second multi-channel input data into the second plaque segmentation model to obtain a second plaque recognition result of the target cross-sectional image;

[0106] A plaque recognition result of the target cross-sectional image is obtained according to the first plaque recognition result and the second plaque recognition result.

[0107] In one embodiment, the segmentation module 303 is specifically configured to:

[0108] The first plaque recognition result and the second plaque recognition result are taken as a union to obtain the plaque recognition result of the target cross-sectional image.

[0109] The specific definition of the plaque identification device can be found in the definition of the plaque identification method above and will not be repeated here. Each module in the aforementioned plaque identification device may be implemented in whole or in part through software, hardware, or a combination thereof. Each of the aforementioned modules may be embedded in or independent of a processor in a computer device in hardware form, or may be stored in a computer device memory in software form, so that the processor can call and execute the corresponding operations of each module.

[0110] In one embodiment, a computer device is provided, whose internal structure diagram can be as follows: Figure 4 As shown. The computer device includes a processor, a memory, and a network interface connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used to store data. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a plaque recognition method is implemented.

[0111] Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0112] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0113] Acquire a cross-sectional image set corresponding to a target cross-sectional image, where the cross-sectional image set includes multiple cross-sectional images, and the target cross-sectional image is one of the multiple cross-sectional images;

[0114] Performing histogram equalization processing on each cross-sectional image in the cross-sectional image set to obtain an equalized image corresponding to each cross-sectional image;

[0115] Multiple cross-sectional images and the equalized images corresponding to each cross-sectional image are input into the trained plaque segmentation model to obtain the plaque recognition result of the target cross-sectional image.

[0116] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0117] Obtain a target cross-sectional image from a centerline point on the centerline of the blood vessel and a cross-sectional image perpendicular to the centerline of the blood vessel, including the centerline point in the three-dimensional image;

[0118] Determine multiple candidate center points on both sides of the center point on the blood vessel center line, and obtain a cross-sectional image corresponding to each candidate center point;

[0119] A set of cross-sectional images corresponding to the target cross-sectional image is obtained according to the target cross-sectional image and the cross-sectional images corresponding to each candidate center point.

[0120] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0121] Taking the centerline point as the starting point, multiple candidate center points are determined on both sides of the centerline point on the centerline of the blood vessel based on a preset step size.

[0122] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0123] Identify and process the blood vessel wall in the target cross-sectional image to obtain a vessel wall annotation map;

[0124] The pipe wall annotation map, multiple cross-sectional images and the equalized images corresponding to each cross-sectional image are input into the trained plaque segmentation model to obtain the plaque recognition result of the target cross-sectional image.

[0125] In one embodiment, the plaque segmentation model includes a first plaque segmentation model, and when the processor executes the computer program, the processor further implements the following steps:

[0126] After normalizing the multiple cross-sectional images and the equalized images corresponding to the cross-sectional images, the pipe wall annotation image, the multiple cross-sectional images, and the equalized images corresponding to the cross-sectional images are merged to obtain first multi-channel input data;

[0127] The first multi-channel input data is input into a first plaque segmentation model to obtain a first plaque recognition result of the target cross-sectional image.

[0128] In one embodiment, the plaque segmentation model further includes a plaque recognition model and a second plaque segmentation model, and when the processor executes the computer program, the processor further implements the following steps:

[0129] After normalizing the multiple cross-sectional images, the pipe wall annotation image and the multiple cross-sectional images are merged to obtain second multi-channel input data;

[0130] inputting the second multi-channel input data into the plaque recognition model to obtain plaque types of plaques in the target cross-sectional image;

[0131] If the plaque type of the plaque in the target cross-sectional image is a preset type, inputting the second multi-channel input data into the second plaque segmentation model to obtain a second plaque recognition result of the target cross-sectional image;

[0132] A plaque recognition result of the target cross-sectional image is obtained according to the first plaque recognition result and the second plaque recognition result.

[0133] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0134] The first plaque recognition result and the second plaque recognition result are taken as a union to obtain the plaque recognition result of the target cross-sectional image.

[0135] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0136] Acquire a cross-sectional image set corresponding to a target cross-sectional image, where the cross-sectional image set includes multiple cross-sectional images, and the target cross-sectional image is one of the multiple cross-sectional images;

[0137] Performing histogram equalization processing on each cross-sectional image in the cross-sectional image set to obtain an equalized image corresponding to each cross-sectional image;

[0138] Multiple cross-sectional images and the equalized images corresponding to each cross-sectional image are input into the trained plaque segmentation model to obtain the plaque recognition result of the target cross-sectional image.

[0139] In one embodiment, when the computer program is executed by the processor, the following steps are further implemented: obtaining a target cross-sectional image from a centerline point on the centerline of the blood vessel and from a cross-sectional image perpendicular to the centerline of the blood vessel in the three-dimensional image including the centerline point;

[0140] Determine multiple candidate center points on both sides of the center point on the blood vessel center line, and obtain a cross-sectional image corresponding to each candidate center point;

[0141] A set of cross-sectional images corresponding to the target cross-sectional image is obtained according to the target cross-sectional image and the cross-sectional images corresponding to each candidate center point.

[0142] In one embodiment, when the computer program is executed by the processor, the following steps are further implemented: starting from the centerline point, multiple candidate center points are determined on both sides of the centerline point on the blood vessel centerline based on a preset step size.

[0143] In one embodiment, when the computer program is executed by the processor, the following steps are further implemented: identifying the blood vessel wall in the target cross-sectional image to obtain a vessel wall annotation image;

[0144] The pipe wall annotation map, multiple cross-sectional images and the equalized images corresponding to each cross-sectional image are input into the trained plaque segmentation model to obtain the plaque recognition result of the target cross-sectional image.

[0145] In one embodiment, the plaque segmentation model includes a first plaque segmentation model, and when the computer program is executed by a processor, the computer program further implements the following steps:

[0146] After normalizing the multiple cross-sectional images and the equalized images corresponding to the cross-sectional images, the pipe wall annotation image, the multiple cross-sectional images, and the equalized images corresponding to the cross-sectional images are merged to obtain first multi-channel input data;

[0147] The first multi-channel input data is input into a first plaque segmentation model to obtain a first plaque recognition result of the target cross-sectional image.

[0148] In one embodiment, the plaque segmentation model further includes a plaque recognition model and a second plaque segmentation model. When the computer program is executed by a processor, the following steps are further implemented:

[0149] After normalizing the multiple cross-sectional images, the pipe wall annotation image and the multiple cross-sectional images are merged to obtain second multi-channel input data;

[0150] inputting the second multi-channel input data into the plaque recognition model to obtain plaque types of plaques in the target cross-sectional image;

[0151] If the plaque type of the plaque in the target cross-sectional image is a preset type, inputting the second multi-channel input data into the second plaque segmentation model to obtain a second plaque recognition result of the target cross-sectional image;

[0152] A plaque recognition result of the target cross-sectional image is obtained according to the first plaque recognition result and the second plaque recognition result.

[0153] In one embodiment, when the computer program is executed by the processor, the following steps are further implemented: taking the union of the first plaque recognition result and the second plaque recognition result to obtain the plaque recognition result of the target cross-sectional image.

[0154] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory, etc. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0155] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0156] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art could make various modifications and improvements without departing from the spirit of the present application, all of which fall within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be determined by the appended claims.

Claims

1. A plaque identification method, characterized in that: The method comprises: Acquire a cross-sectional image set corresponding to a target cross-sectional image, wherein the cross-sectional image set includes a plurality of cross-sectional images, and the target cross-sectional image is one of the plurality of cross-sectional images; performing histogram equalization processing on each of the cross-sectional images in the cross-sectional image set to obtain an equalized image corresponding to each of the cross-sectional images; Inputting the plurality of cross-sectional images and the equalized images corresponding to the cross-sectional images into a trained plaque segmentation model to obtain a plaque recognition result of the target cross-sectional image; The method also includes: identifying and processing the blood vessel wall in the target cross-sectional image to obtain a vessel wall annotation map; inputting the multiple cross-sectional images and the equalized images corresponding to each of the cross-sectional images into a trained plaque segmentation model to obtain a plaque recognition result for the target cross-sectional image, including: inputting the vessel wall annotation map, the multiple cross-sectional images, and the equalized images corresponding to each of the cross-sectional images into the trained plaque segmentation model to obtain a plaque recognition result for the target cross-sectional image.

2. The method according to claim 1, characterized in that The acquiring of a set of cross-sectional images corresponding to the target cross-sectional image comprises: Acquiring a plurality of centerline points on the centerline of the blood vessel, and for each centerline point, intercepting a cross-sectional image perpendicular to the centerline of the blood vessel and including the centerline point from the three-dimensional image to obtain the target cross-sectional image; Determine a plurality of candidate center points on both sides of the center point on the center line of the blood vessel, and obtain a cross-sectional image corresponding to each candidate center point; A set of cross-sectional images corresponding to the target cross-sectional image is obtained according to the target cross-sectional image and the cross-sectional images corresponding to the candidate center points.

3. The method according to claim 2, characterized in that Determining a plurality of candidate center points on both sides of the center point on the blood vessel center line includes: Taking the centerline point as a starting point, multiple candidate center points are determined on the centerline of the blood vessel toward both sides of the centerline point based on a preset step size.

4. The method according to claim 1, wherein The plaque segmentation model includes a first plaque segmentation model, and the step of inputting the pipe wall annotation map, the multiple cross-sectional images, and the equalized images corresponding to the cross-sectional images into the trained plaque segmentation model to obtain a plaque recognition result of the target cross-sectional image includes: After normalizing the multiple cross-sectional images and the equalized images corresponding to the cross-sectional images, the pipe wall annotation image, the multiple cross-sectional images, and the equalized images corresponding to the cross-sectional images are merged to obtain first multi-channel input data; The first multi-channel input data is input into the first plaque segmentation model to obtain a first plaque recognition result of the target cross-sectional image.

5. The method according to claim 4, characterized in that The plaque segmentation model further includes a plaque recognition model and a second plaque segmentation model. The process of inputting the pipe wall annotation map, the multiple cross-sectional images, and the equalized images corresponding to the cross-sectional images into the trained plaque segmentation model to obtain a plaque recognition result of the target cross-sectional image includes: After normalizing the multiple cross-sectional images, the pipe wall annotation image and the multiple cross-sectional images are merged to obtain second multi-channel input data; inputting the second multi-channel input data into the plaque recognition model to obtain the plaque type of the plaque in the target cross-sectional image; If the plaque type of the plaque in the target cross-sectional image is a preset type, inputting the second multi-channel input data into the second plaque segmentation model to obtain a second plaque recognition result of the target cross-sectional image; A plaque recognition result of the target cross-sectional image is obtained according to the first plaque recognition result and the second plaque recognition result.

6. The method according to claim 5, characterized in that Obtaining a plaque recognition result of the target cross-sectional image according to the first plaque recognition result and the second plaque recognition result includes: The first plaque recognition result and the second plaque recognition result are taken as a union to obtain a plaque recognition result of the target cross-sectional image.

7. A plaque identification device, characterized in that: The device comprises: an acquisition module, configured to acquire a cross-sectional image set corresponding to a target cross-sectional image, wherein the cross-sectional image set includes a plurality of cross-sectional images, and the target cross-sectional image is one of the plurality of cross-sectional images; a processing module, configured to perform histogram equalization processing on each of the cross-sectional images in the cross-sectional image set to obtain an equalized image corresponding to each of the cross-sectional images; a segmentation module, configured to input the plurality of cross-sectional images and the equalized images corresponding to the cross-sectional images into a trained plaque segmentation model to obtain a plaque recognition result of the target cross-sectional image; The segmentation module is further used to identify and process the blood vessel wall in the target cross-sectional image to obtain a vessel wall annotation map; the step of inputting the multiple cross-sectional images and the equalized images corresponding to each of the cross-sectional images into a trained plaque segmentation model to obtain a plaque recognition result for the target cross-sectional image includes: inputting the vessel wall annotation map, the multiple cross-sectional images, and the equalized images corresponding to each of the cross-sectional images into a trained plaque segmentation model to obtain a plaque recognition result for the target cross-sectional image.

8. The device according to claim 7, characterized in that The acquisition module is specifically used to acquire multiple centerline points on the centerline of the blood vessel. For each centerline point, a cross-sectional image including the centerline point and perpendicular to the centerline of the blood vessel is intercepted from the three-dimensional stereo image to obtain the target cross-sectional image; multiple candidate center points are determined on both sides of the centerline point on the centerline of the blood vessel, and a cross-sectional image corresponding to each candidate center point is acquired; and a cross-sectional image set corresponding to the target cross-sectional image is obtained based on the target cross-sectional image and the cross-sectional images corresponding to each candidate center point.

9. A computer device, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the method according to any one of claims 1 to 6 is implemented.

10. A computer-readable storage medium, characterized in that A computer program is stored thereon, and when the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images