Image Processing Method, System, Electronic Device, and Storage Medium

Abstract

The present application discloses an image processing method, system, electronic device and storage medium. Among them, the image processing method includes: inputting a first target magnetic resonance image into a brainstem generation model for modeling processing to obtain a three-dimensional brainstem image, and obtaining a target brainstem image according to the three-dimensional brainstem image; inputting the target brainstem image into a target boundary point detection model for detection processing to obtain boundary point coordinates; obtaining the midbrain area and pons area according to the boundary point coordinates, target point coordinates and target brainstem image; inputting a second target magnetic resonance image into an endpoint detection model to obtain endpoint coordinates, and then calculating the middle cerebellar peduncle data according to the endpoint coordinates; obtaining the superior cerebellar peduncle data according to the second magnetic resonance image, and finally obtaining target data according to the midbrain area, pons area, middle cerebellar peduncle data and superior cerebellar peduncle data. The image processing method of the embodiments of the present application can improve the measurement efficiency and measurement accuracy.

Description

Technical Field

[0001] The present application relates to the technical field of image processing, and in particular, to an image processing method, system, electronic device, and storage medium. Background Art

[0002] Currently, the Magnetic Resonance Parkinson Index (MRPI) is an index used clinically to determine whether a patient has Parkinson's disease or Parkinson's plus syndrome. Obtaining the MRPI index requires obtaining various parameters from magnetic resonance images (MRI), including: the area of the midbrain, the area of the pons, the width of the middle cerebellar peduncle, and the width of the superior cerebellar peduncle.

[0003] In the related art, the above parameters are measured by the Otsu threshold segmentation algorithm. However, during the measurement process of the Otsu threshold segmentation algorithm, it is necessary for the measurement personnel to manually annotate and screen the magnetic resonance images, resulting in a cumbersome measurement process. At the same time, the accuracy of the measured parameters is related to the professional level of the measurement personnel, resulting in low accuracy and unreliability of the measurement results. Summary of the Invention

[0004] The present application aims to at least solve one of the technical problems existing in the prior art. For this reason, the present application proposes an image processing method that can improve the measurement efficiency and measurement accuracy.

[0005] The present application also proposes an image processing system, an electronic device applying the above image processing method, and a computer-readable storage medium applying the above image processing method.

[0006] The image processing method according to the first aspect embodiment of the present application includes:

[0007] Obtain a first magnetic resonance image and a second magnetic resonance image; wherein, the first magnetic resonance image is used to represent the sagittal plane image of the brain, the second magnetic resonance image is used to represent the coronal plane image of the brain, the first magnetic resonance image includes: a first target magnetic resonance image and a second target magnetic resonance image, the first target magnetic resonance image is an image containing the brainstem region, the second target magnetic resonance image is an image containing the middle cerebellar peduncle region, and the second magnetic resonance image is an image containing the superior cerebellar peduncle region;

[0008] Input the first target magnetic resonance image into a preset brainstem generation model for modeling processing to obtain a three-dimensional brainstem image;

[0009] Perform cutting processing on the three-dimensional brainstem image to obtain a brainstem cut image and a first coordinate; wherein, the first coordinate is used to represent the coordinate of the bottom end of the quadrigeminal region in the brainstem cut image;

[0010] Obtain the area of the brainstem cutting image, and use the brainstem cutting image corresponding to the minimum value of the area as the target brainstem image; wherein, use the first coordinate of the target brainstem image as the target point coordinate;

[0011] Input the target brainstem image into a preset target boundary point detection model for detection processing to obtain boundary point coordinates;

[0012] Obtain the midbrain area based on the boundary point coordinates, the target point coordinates, and the target brainstem image, and obtain the pons area based on the boundary point coordinates, the target point coordinates, and the target brainstem image;

[0013] Input the second target magnetic resonance image into a preset target endpoint detection model for detection processing to obtain target endpoint coordinates, and obtain the middle cerebellar peduncle data based on the target endpoint coordinates;

[0014] Obtain the superior cerebellar peduncle data based on the second magnetic resonance image;

[0015] Obtain target data based on the midbrain area, the pons area, the middle cerebellar peduncle data, and the superior cerebellar peduncle data.

[0016] According to the image processing method of the embodiments of the present application, it has at least the following beneficial effects: Obtain a first magnetic resonance image representing the sagittal plane image of the brain and a second magnetic resonance image representing the coronal plane image of the brain. Among them, the first magnetic resonance image includes a first target magnetic resonance image and a second target magnetic resonance image. The first target magnetic resonance image contains the brainstem part, and the second target magnetic resonance image contains the middle cerebellar peduncle part. Input the first target magnetic resonance image into a brainstem generation model to obtain a three-dimensional brainstem image, and perform cutting processing on the three-dimensional brainstem image to obtain a brainstem cutting image and a first coordinate. Among them, take the brainstem cutting image corresponding to the minimum value of the area of the brainstem cutting image as the target brainstem image, and use the first coordinate in the target brainstem image as the target point coordinate. Input the target brainstem image into a boundary point detection model to obtain boundary point coordinates, and calculate the midbrain area and the pons area through the boundary point coordinates, the target point coordinates, and the target brainstem image. Input the second target magnetic resonance image into an endpoint detection model to obtain endpoint coordinates, and then calculate the middle cerebellar peduncle data according to the endpoint coordinates. Obtain the superior cerebellar peduncle data according to the second magnetic resonance image, and finally obtain the target data according to the midbrain area, the pons area, the middle cerebellar peduncle data, and the superior cerebellar peduncle data, thereby realizing the automatic calculation of the midbrain area, the pons area, the middle cerebellar peduncle data, and the superior cerebellar peduncle data. The image processing method of this embodiment does not require manual annotation and screening of magnetic resonance images, thereby effectively shortening the measurement process and improving the measurement efficiency. At the same time, the midbrain area data, the pons area data, and the middle cerebellar peduncle data are all calculated by using a network model, improving the accuracy of the measurement results.

[0017] According to some embodiments of the present application, the boundary point coordinates include: first sub-boundary point coordinates and second sub-boundary point coordinates;

[0018] Obtaining the midbrain area based on the boundary point coordinates, the target point coordinates, and the target brainstem image, and obtaining the pons area based on the boundary point coordinates, the target point coordinates, and the target brainstem image, includes:

[0019] Obtaining a first boundary line based on the first sub-boundary point coordinates and the target point coordinates;

[0020] Obtaining a midbrain region based on the first boundary line and the target brainstem image, and obtaining the midbrain area based on the midbrain region;

[0021] Obtaining a second boundary line based on the second sub-boundary point coordinates and the first boundary line; wherein, the second boundary line is parallel to the first boundary line;

[0022] Obtaining a pons region based on the first boundary line, the second boundary line, and the target brainstem image, and obtaining the pons area based on the pons region.

[0023] According to some embodiments of the present application, before inputting the target brainstem image into a preset target boundary point detection model for detection processing to obtain boundary point coordinates, the method further includes training the target boundary point detection model, specifically including:

[0024] Obtaining a sample brainstem image and sample boundary point coordinates of the sample brainstem image;

[0025] Inputting the sample brainstem image into a preset original boundary point detection model for detection processing to obtain original boundary point coordinates;

[0026] Adjusting the parameters of the original boundary point detection model according to the original boundary point coordinates and the sample boundary point coordinates to obtain the target boundary point detection model.

[0027] According to some embodiments of the present application, the inputting the second target magnetic resonance image into a preset target endpoint detection model for detection processing to obtain target endpoint coordinates includes:

[0028] Inputting the second target magnetic resonance image into the target endpoint detection model for detection processing to obtain a target endpoint probability distribution image;

[0029] Obtaining the target endpoint coordinates according to the target endpoint probability distribution image.

[0030] According to some embodiments of the present application, the endpoint probability distribution image includes a first original coordinate and a first distribution probability of the first original coordinate;

[0031] Obtaining the target endpoint coordinates according to the target endpoint probability distribution image includes:

[0032] Comparing the first distribution probabilities to obtain a maximum first probability;

[0033] Taking the first original coordinate corresponding to the maximum first probability as the target endpoint coordinate.

[0034] According to some embodiments of the present application, before inputting the second target magnetic resonance image into the target endpoint detection model for detection processing to obtain the target endpoint probability distribution image, the method further includes: training the target endpoint detection model, specifically including:

[0035] Obtaining a sample middle cerebellar peduncle image and sample endpoint coordinates of the sample middle cerebellar peduncle image;

[0036] Inputting the sample middle cerebellar peduncle image into a preset original endpoint detection model for detection processing to obtain an original endpoint probability distribution image; wherein, the original endpoint probability distribution image includes a second original coordinate and a second distribution probability of the second original coordinate;

[0037] Comparing the second distribution probabilities to obtain a maximum second probability;

[0038] Taking the second original coordinate corresponding to the maximum second probability as the original endpoint coordinate;

[0039] Adjusting the parameters of the original endpoint detection model according to the original endpoint coordinates and the sample endpoint coordinates to obtain the target endpoint detection model.

[0040] According to some embodiments of the present application, obtaining the superior cerebellar peduncle data from the second magnetic resonance image includes:

[0041] Obtaining a centroid coordinate and a bisector from the second magnetic resonance image; wherein, the centroid coordinate is used to represent the coordinate of the centroid of the superior cerebellar peduncle;

[0042] Obtaining a width line according to the centroid coordinate and the bisector;

[0043] Obtaining the superior cerebellar peduncle data according to the width line and the second magnetic resonance image.

[0044] An image processing system according to the second aspect embodiment of the present application includes:

[0045] An image acquisition module, which is used to acquire a first magnetic resonance image and a second magnetic resonance image; wherein, the first magnetic resonance image is used to represent the sagittal plane image of the brain, and the second magnetic resonance image is used to represent the coronal plane image of the brain. The first magnetic resonance image includes: a first target magnetic resonance image and a second target magnetic resonance image. The first target magnetic resonance image is an image containing the brainstem region, and the second target magnetic resonance image is an image containing the middle cerebellar peduncle region. The second magnetic resonance image is an image containing the superior cerebellar peduncle region;

[0046] A brainstem image processing module, which is used to input the first target magnetic resonance image into a preset brainstem generation model for modeling processing to obtain a three-dimensional brainstem image; perform cutting processing on the three-dimensional brainstem image to obtain a brainstem cut image and a first coordinate; wherein, the first coordinate is used to represent the coordinate of the bottom end of the quadrigeminal region in the brainstem cut image; obtain the area of the brainstem cut image, and use the brainstem cut image corresponding to the minimum value of the area as the target brainstem image; wherein, use the first coordinate of the target brainstem image as the target point coordinate;

[0047] An area acquisition module, which is used to input the target brainstem image into a preset target boundary point detection model for detection processing to obtain boundary point coordinates; obtain the midbrain area according to the boundary point coordinates, the target point coordinates, and the target brainstem image, and obtain the pons area according to the boundary point coordinates, the target point coordinates, and the target brainstem image;

[0048] A middle cerebellar peduncle data acquisition module, which is used to input the second target magnetic resonance image into a preset target endpoint detection model for detection processing to obtain target endpoint coordinates, and obtain middle cerebellar peduncle data according to the target endpoint coordinates;

[0049] A superior cerebellar peduncle data acquisition module, which is used to obtain superior cerebellar peduncle data according to the second magnetic resonance image;

[0050] A target data calculation module, which is used to obtain target data according to the midbrain area, the pons area, the middle cerebellar peduncle data, and the superior cerebellar peduncle data.

[0051] The image processing system according to the embodiments of the present application has at least the following beneficial effects: By adopting the above image processing method, the automatic calculation of the midbrain area, the pons area, the middle cerebellar peduncle data, and the superior cerebellar peduncle data is realized, and there is no need to manually annotate and screen magnetic resonance images, thereby effectively shortening the measurement process and improving the measurement efficiency.

[0052] An electronic device according to an embodiment of the third aspect of the present application includes:

[0053] At least one memory;

[0054] At least one processor;

[0055] At least one computer program;

[0056] The computer program is stored in the memory, and the processor executes the at least one computer program to implement the image processing method according to the embodiment of the first aspect above.

[0057] A computer-readable storage medium according to an embodiment of the fourth aspect of the present application includes:

[0058] The computer-readable storage medium stores computer-executable instructions for causing a computer to execute the image processing method according to the embodiment of the first aspect above.

[0059] Additional aspects and advantages of the present application will be given in part in the following description, become apparent in part from the following description, or be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] The present application will be further described below in conjunction with the drawings and embodiments, where:

[0061] Figure 1 is a flowchart of the image processing method provided by the embodiment of the present application;

[0062] Figure 2 is a schematic diagram of the target brainstem image in the image processing method provided by the embodiment of the present application;

[0063] Figure 3 is a schematic diagram of the middle cerebellar peduncle image in the image processing method provided by the embodiment of the present application;

[0064] Figure 4 is a schematic diagram of the superior cerebellar peduncle image in the image processing method provided by the embodiment of the present application;

[0065] Figure 5 is a flowchart of the specific method for training the target boundary point detection model in the embodiment of the present application;

[0066] Figure 6 is Figure 1 a flowchart of the specific method of step S600 in

[0067] Figure 7 is Figure 1 a flowchart of the specific method of step S700 in

[0068] Figure 8 isFigure 7 Flow chart of the specific method in step S720;

[0069] Figure 9 Schematic diagram of the endpoint probability distribution image in the embodiment of the present application;

[0070] Figure 10 Flow chart of the specific method for training the target endpoint detection model in the embodiment of the present application;

[0071] Figure 11 Schematic diagram of the original endpoint probability distribution image in the embodiment of the present application;

[0072] Figure 12 is Figure 1 Flow chart of the specific method in step S800;

[0073] Figure 13 Block diagram of the image processing system provided in the embodiment of the present application.

[0074] Figure 14 Schematic diagram of the hardware structure of the electronic device provided in the embodiment of the present application.

[0075] Reference numerals:

[0076] Quadrigeminal region 100, midbrain region 200, pons region 300, left superior cerebellar peduncle region 400, right superior cerebellar peduncle region 500, image acquisition module 610, brainstem image processing module 620, area acquisition module 630, middle cerebellar peduncle data acquisition module 640, superior cerebellar peduncle data acquisition module 650, target data calculation module 660, processor 710, memory 720, input / output interface 730, communication interface 740, bus 750. Detailed implementation manners

[0077] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, and are not used to limit the present application.

[0078] It should be noted that although the functional modules are divided in the device schematic diagram and the logical order is shown in the flow chart, in some cases, the steps shown or described may be executed in a different order from the module division in the device or the order in the flow chart. Terms such as "first" and "second" in the specification, claims and the above-mentioned drawings are used to distinguish similar objects, and do not have to be used to describe a specific order or sequence.

[0079] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terms used herein are for the purpose of describing embodiments of this application only and are not intended to limit this application.

[0080] First, the following are the explanations of several terms involved in this application:

[0081] Artificial Intelligence (AI): It is a new technical science that studies and develops theories, methods, technologies, and application systems for simulating, extending, and expanding human intelligence; artificial intelligence is a branch of computer science. It attempts to understand the essence of intelligence and produce a new intelligent machine that can respond in a way similar to human intelligence. The research in this field includes robots, speech recognition, image recognition, natural language processing, and expert systems, etc. Artificial intelligence can simulate the information process of human consciousness and thinking. It is also a theory, method, technology, and application system that uses a digital computer or a machine controlled by a digital computer to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to obtain the best results.

[0082] Machine Learning: It is a special study on how a computer simulates or realizes human learning behavior to acquire new knowledge or skills and reorganize the existing knowledge structure to continuously improve its own performance. Machine learning is the core of artificial intelligence and the fundamental way to make a computer intelligent. In machine learning, Deep Learning (DL) is to learn the internal laws and representation levels of sample data, and the information obtained in these learning processes is very helpful for the interpretation of data such as text, images, and sounds. Its ultimate goal is to enable the machine to have the ability of analysis and learning like a human being and be able to recognize data such as text, images, and sounds. Deep learning is a complex machine learning algorithm, and the effects achieved in speech and image recognition far exceed those of previous related technologies.

[0083] UNet network model: It is used for image segmentation, and its most significant feature is the U-shaped structure and skip connections. After the image is input into the UNet network model, through four downsampling operations in the convolutional layer, deep regional features are obtained. Then, deconvolution upsampling operations are performed on the deep regional features. Finally, the restored image has the same size as the input image, so as to detect each pixel of the image and at the same time be able to retain the three-dimensional spatial information of the image. Among them, the network model structure with downsampling operations first and then upsampling operations is the U-shaped structure. Skip connections ensure that the finally restored image can fuse more shallow features, so that the image segmentation result can be more precise.

[0084] HRNet network model: The multi-resolution subnetworks are connected to the low-resolution subnetwork in a parallel connection manner. This model takes the high-resolution subnetwork as the first stage, gradually adds high-resolution to low-resolution subnetworks to form more stages, and connects the multi-resolution subnetworks in parallel. The HRNet network model performs multi-scale repeated fusion by repeatedly exchanging information on the parallel multi-resolution subnetworks, and estimates key points through the high-resolution representation output by the network.

[0085] Otsu threshold segmentation algorithm: This algorithm separates the image into foreground and background parts according to the gray value by finding a gray value as the threshold, so as to isolate the area to be processed from the image.

[0086] MNI (Montreal Neurological Institute) space: It is a coordinate system established based on a series of magnetic resonance images of normal human brains. After obtaining the magnetic resonance images, different magnetic resonance images are in different original spaces, and the dimensions, origins, voxel sizes, etc. of the images in this space are different, and the different images are not comparable, and any features calculated cannot be statistically analyzed or used for machine learning. Registering all the magnetic resonance images to be processed into the MNI space makes the dimensions, origins, voxel sizes, etc. of all images unified, so that statistical analysis can be performed on all magnetic resonance images.

[0087] Magnetic Resonance Parkinson Index (MRPI): It is an index used to clinically determine whether a patient has Parkinson's disease or Parkinson's plus syndrome. Its calculation formula is:

[0088]

[0089] Among them, S midbrain represents the midbrain area, S pon represents the pons area, MCP represents the width of the bilateral middle cerebellar peduncles, and SCP represents the width of the bilateral superior cerebellar peduncles.

[0090] In the related art, the above midbrain area, pons area, width of the middle cerebellar peduncles, and width of the superior cerebellar peduncles are measured by the Otsu threshold segmentation algorithm. However, during the measurement process of the Otsu threshold segmentation algorithm, the measurement personnel need to manually annotate and screen the magnetic resonance images, which makes the measurement process cumbersome. At the same time, the accuracy of the measured parameters is related to the professional level of the measurement personnel, resulting in low accuracy and unreliability of the measurement results.

[0091] Based on this, embodiments of the present disclosure provide an image processing method, system, electronic device, and storage medium, which can automatically obtain midbrain area data, pons area data, middle cerebellar peduncle data, and superior cerebellar peduncle data, without the need for manual annotation and screening of magnetic resonance images, thus effectively shortening the measurement process. At the same time, the midbrain area data, pons area data, and middle cerebellar peduncle data are all obtained by using a network model, improving the accuracy of the measurement results.

[0092] As Figure 1 shown, embodiments of the present application provide an image processing method, which includes but is not limited to steps S100 to S900. The following is a detailed introduction to these nine steps.

[0093] Step S100: Obtain a first magnetic resonance image and a second magnetic resonance image; wherein, the first magnetic resonance image is used to represent the sagittal plane image of the brain, and the second magnetic resonance image is used to represent the coronal plane image of the brain. The first magnetic resonance image includes: a first target magnetic resonance image and a second target magnetic resonance image. The first target magnetic resonance image is an image containing the brainstem region, and the second target magnetic resonance image is an image containing the middle cerebellar peduncle region.

[0094] Specifically, perform multi-layer scanning operations on the brain in the sagittal plane direction to obtain multiple first magnetic resonance images. Among them, the multiple first magnetic resonance images cover all brain tissue regions to be detected, such as the brainstem region, cerebellar region, etc. The image containing the brainstem region is used as the first target magnetic resonance image, and the image containing the middle cerebellar peduncle region is used as the second target magnetic resonance image.

[0095] The number of first magnetic resonance images obtained can be no less than 40, that is, at least 40 layers of scanning need to be performed on the brain during the scanning operation. The scanning intensity of the scanning operation can be 1.5T or 3T. The obtained first magnetic resonance images are registered to the MNI space, and 6-degree-of-freedom rigid registration can be used in the registration to ensure that data with different scanning parameters and different scanning devices can be uniformly transformed into the same MNI space. The coordinates in the following embodiments are the coordinates in this MNI space.

[0096] In the above first magnetic resonance images, assuming that the number of images containing the brainstem region is 10, any number of them can be selected as the first target magnetic resonance image. Assuming that the number of images containing the middle cerebellar peduncle region is 20, with 10 images of the left middle cerebellar peduncle region and 10 images of the right middle cerebellar peduncle region. Among them, any number of images containing the middle cerebellar peduncle region can be selected as the second target magnetic resonance image. It should be noted that the specific number of images selected as the first target magnetic resonance image or the second target magnetic resonance image can be adaptively adjusted according to actual needs.

[0097] Step S200: Input the first target magnetic resonance image into a preset brainstem generation model for modeling to obtain a three-dimensional brainstem image.

[0098] Specifically, the preset brainstem generation model can be a UNet network model. After the first target magnetic resonance image is input into the UNet network model, the UNet network model performs modeling based on the first target magnetic resonance image to obtain a three-dimensional brainstem image. Among them, the three-dimensional brainstem image can specifically be a 3D structure image of the brainstem.

[0099] Step S300: Perform cutting processing on the three-dimensional brainstem image to obtain a brainstem cut image and a first coordinate; where the first coordinate is used to represent the coordinate of the bottom end of the quadrigeminal region in the brainstem cut image.

[0100] Specifically, perform cutting processing on the three-dimensional brainstem image from the sagittal plane direction to obtain a brainstem cut image. For example, the three-dimensional brainstem image can be cut 10 times to obtain 10 brainstem cut images. Each brainstem cut image contains the quadrigeminal region. Referring to Figure 2 , the lowermost end of the quadrigeminal region 100 is the bottom end. The bottom end of the quadrigeminal region 100 is a uniquely determined point, and the coordinate of the point in the corresponding brainstem cut image is the first coordinate.

[0101] Step S400: Obtain the area of the brainstem cut image, and use the brainstem cut image corresponding to the minimum value of the area as the target brainstem image; where the first coordinate of the target brainstem image is used as the target point coordinate.

[0102] Obtain the area of each brainstem cut image, and use the brainstem cut image corresponding to the minimum area value as the target brainstem image. In anatomy, the median sagittal plane is located in the middle layer between the left and right cerebral hemispheres, and the cross-sectional area of the brainstem on this plane is the smallest. Therefore, the target brainstem image is the image of the cross-section of the brainstem on the median sagittal plane. After determining the target brainstem image, use the first coordinate in the target brainstem image as the target point coordinate. For example, referring to Figure 2 , point a at the bottom end of the quadrigeminal region 100 is the target point. Therefore, the coordinate of point a is the target point coordinate.

[0103] Step S500: Input the target brainstem image into a preset target boundary point detection model for detection processing to obtain boundary point coordinates.

[0104] Specifically, the preset target boundary point detection model can be an HRNet network model. The HRNet network model detects boundary points based on the input target brainstem image and outputs boundary point coordinates.

[0105] Step S600: Obtain the midbrain area based on the boundary point coordinates, target point coordinates, and target brainstem image, and obtain the pons area based on the boundary point coordinates, target point coordinates, and target brainstem image.

[0106] Specifically, based on the boundary point coordinates and target point coordinates, the midbrain area and pons area in the target brainstem image can be determined.

[0107] Step S700: Input the second target magnetic resonance image into a preset target endpoint detection model for detection processing to obtain target endpoint coordinates, and obtain the middle cerebellar peduncle data based on the endpoint coordinates.

[0108] Specifically, the preset target endpoint detection model can be an HRNet network model, and the middle cerebellar peduncle data includes the above-mentioned middle cerebellar peduncle width. The target endpoint detection model performs detection processing on the second target magnetic resonance image to obtain target endpoint coordinates. Among them, the target endpoints are the endpoints for obtaining the middle cerebellar peduncle width. After obtaining the target endpoint coordinates, the middle cerebellar peduncle width can be obtained according to the target endpoint coordinates. For example, referring to Figure 3 , point m and point n are the target endpoints, and Lmcp is the middle cerebellar peduncle width.

[0109] Step S800: Obtain the superior cerebellar peduncle data based on the second magnetic resonance image.

[0110] Specifically, the superior cerebellar peduncle data includes the above-mentioned superior cerebellar peduncle width. Referring to Figure 4 , select the sectional images of the left superior cerebellar peduncle region 400 and the right superior cerebellar peduncle region 500 from the second magnetic resonance image. Obtain the width of the left superior cerebellar peduncle (i.e., Lscp in Figure 4 ) according to the left superior cerebellar peduncle region 400 in this image, obtain the width of the right superior cerebellar peduncle according to the right superior cerebellar peduncle region 500 in this image, calculate the average value of the width of the left superior cerebellar peduncle and the width of the right superior cerebellar peduncle, and take this average value as the superior cerebellar peduncle width. Among them, the left superior cerebellar peduncle region 400 and the right superior cerebellar peduncle region 500 in the second magnetic resonance image can be separated by using the Otsu threshold segmentation algorithm.

[0111] Step S900: Obtain the target data based on the midbrain area, pons area, middle cerebellar peduncle data, and superior cerebellar peduncle data.

[0112] Specifically, the target data is the magnetic resonance Parkinson's index. Substitute the midbrain area, pons area, middle cerebellar peduncle data, and superior cerebellar peduncle data obtained in the above steps S100 to S800 into Equation (1), and the magnetic resonance Parkinson's index can be obtained. The following Table 1 shows the comparison of the measurement performance between the embodiments of the present invention and the related art through the Otsu threshold segmentation algorithm.

[0113] Table 1:

[0114] Relative error Absolute error Standard deviation of absolute error Embodiment of the present invention 18.17％ 0.21 1.30 Otsu threshold segmentation algorithm 32.23％ 0.41 2.07

[0115] As shown in Table 1, the relative error is the difference between the magnetic resonance Parkinson's index obtained by automatic measurement (i.e., the measured value) and the magnetic resonance Parkinson's index measured by professional measurers (i.e., the true value), and the absolute error is the difference between the relative error and the true value. It can be seen from Table 1 that the relative error of the magnetic resonance Parkinson's index obtained by the image processing method of the embodiment of the present invention is lower, and the standard deviation of the absolute error is smaller, indicating that the image processing method of the embodiment of the present invention can improve the accuracy of the measurement result.

[0116] The image processing method proposed in the embodiment of the present application obtains a first magnetic resonance image, inputs the first target magnetic resonance image in the first magnetic resonance image into a preset brainstem generation model for modeling processing to obtain a three-dimensional brainstem image. The three-dimensional brainstem image is subjected to cutting processing to obtain a brainstem cut image and a first coordinate, and the brainstem cut image corresponding to the minimum area of the brainstem cut image is taken as the target brainstem image, and the first coordinate in the target brainstem image is taken as the target point coordinate. The target brainstem image is input into a boundary point detection model to obtain boundary point coordinates, and the midbrain area and pons area are calculated through the boundary point coordinates, the target point coordinates, and the target brainstem image. The second target magnetic resonance image is input into an endpoint detection model to obtain endpoint coordinates, and then the middle cerebellar peduncle data is calculated according to the endpoint coordinates. The superior cerebellar peduncle data is obtained according to the second magnetic resonance image, and finally the target data is obtained according to the midbrain area, the pons area, the middle cerebellar peduncle data, and the superior cerebellar peduncle data, thereby realizing the automatic calculation of the midbrain area, the pons area, the middle cerebellar peduncle data, and the superior cerebellar peduncle data. The image processing method of this embodiment does not require manual annotation and screening of magnetic resonance images, thereby effectively shortening the measurement process and improving the measurement efficiency. At the same time, the midbrain area data, the pons area data, and the middle cerebellar peduncle data are all calculated by using a network model, improving the accuracy of the measurement result.

[0117] As Figure 5 shown, in some embodiments of the present application, before step S500, the image processing method further includes training a target boundary point detection model, specifically including but not limited to steps S510 to S530, and the following is a detailed introduction to these three steps.

[0118] Step S510: Obtain a sample brainstem image and sample boundary point coordinates of the sample brainstem image.

[0119] Specifically, the sample brainstem image is a cross-sectional image of the brainstem on the median sagittal plane. The sample boundary points are obtained by annotation and are used to divide the midbrain region and the pons region from the brainstem image.

[0120] Step S520: Input the sample brainstem image into a preset original boundary point detection model for detection processing to obtain the original boundary point coordinates.

[0121] Specifically, the preset original boundary point detection model can be an HRNet network model. The HRNet network model detects the boundary points based on the input sample brainstem image and outputs the original boundary point coordinates.

[0122] Step S530: Adjust the parameters of the original boundary point detection model according to the original boundary point coordinates and the sample boundary point coordinates to obtain the target boundary point detection model.

[0123] Specifically, compare the original boundary point coordinates with the sample boundary point coordinates to obtain the difference between the boundary point coordinates and the sample boundary point coordinates. Then, adjust the parameters of the original boundary point detection model according to this difference to obtain the target boundary point detection model. The target boundary point detection model trained in this embodiment can more accurately predict the boundary point coordinates.

[0124] In some embodiments of the present application, the boundary point coordinates include: the first sub-boundary point coordinates and the second sub-boundary point coordinates. As Figure 6 shown, step S600 includes but is not limited to steps S610 to S640. These four steps will be introduced in detail below.

[0125] Step S610: Obtain the first boundary line according to the first sub-boundary point coordinates and the target point coordinates.

[0126] Specifically, referring to Figure 2 , the first sub-boundary point is the upper tangent point (point b) of the pons region 300, and the target point is point a at the bottom of the quadrigeminal region 100. After obtaining the boundary point coordinates from the preset target boundary point detection model, the first sub-boundary point coordinates and the target point coordinates are known. The line segment A obtained by connecting the first boundary point b and the target point a is the first boundary line.

[0127] Step S620: Obtain the midbrain region according to the first boundary line and the target brainstem image, and obtain the midbrain area according to the midbrain region.

[0128] Specifically, referring to Figure 2 , the first boundary line A can divide the midbrain region 200 from the target brainstem image.

[0129] Step S630: Obtain the second boundary line according to the second sub-boundary point coordinates and the first boundary line; wherein, the second boundary line is parallel to the first boundary line.

[0130] Specifically, referring to Figure 2, the second boundary point is the lower tangent point (point c) of the pons region 300. After obtaining the first boundary line, draw a line segment B through the second boundary point c as the second boundary line, and the second boundary line is parallel to the first boundary line.

[0131] Step S640: Obtain the pons region according to the first boundary line, the second boundary line, and the target brainstem image, and obtain the pons area according to the pons region.

[0132] Specifically, referring to Figure 2 , the first boundary line A and the second boundary line B can segment the pons region 300 from the target brainstem image, and the pons region 300 is located between the first boundary line A and the second boundary line B. Finally, the midbrain area can be calculated according to the midbrain region 200, and the pons area can be calculated according to the pons region 300. The method of this embodiment determines the first boundary line A and the second boundary line B through the target point coordinates, the first sub-boundary point coordinates, and the second sub-boundary point coordinates, and divides the target brainstem image through the first boundary line A and the second boundary line B to obtain the midbrain region 200 and the pons region 300. Among them, the first sub-boundary point coordinates and the second sub-boundary point coordinates are both the results output by the target boundary point detection model. Therefore, the midbrain region and the pons region divided by the method of this embodiment have high accuracy.

[0133] As Figure 7 shown, in some embodiments of the present application, step S700 includes but is not limited to step S710 and step S720, and these two steps will be introduced in detail below.

[0134] Step S710: Input the second target magnetic resonance image into the target endpoint detection model for detection processing to obtain the target endpoint probability distribution image.

[0135] Specifically, the preset target endpoint detection model can be an HRNet network model. The HRNet network model detects according to the input second target magnetic resonance image and outputs the target endpoint probability distribution image. Among them, the target endpoint probability distribution image contains the distribution probability information of the endpoint coordinates.

[0136] Step S720: Obtain the target endpoint coordinates according to the target endpoint probability distribution image.

[0137] Specifically, since the target endpoint probability distribution image contains the distribution probability information of the endpoint coordinates, the target endpoint coordinates can be screened out from the target endpoint probability distribution image according to the distribution probability information of the endpoint coordinates.

[0138] As Figure 8 , Figure 9As shown, in some embodiments of the present application, the target endpoint probability distribution image includes a first original coordinate and a first distribution probability of the first original coordinate. Step S720 includes, but is not limited to, step S721 and step S722. The following provides a detailed introduction to these two steps.

[0139] Step S721: Compare the first distribution probabilities to obtain the first maximum probability.

[0140] Specifically, the target endpoint probability distribution image can be an image as shown in Figure 9 . The X-axis is the first original coordinate, and the Y-axis is the first distribution probability of the first original coordinate. This distribution probability represents the probability that the endpoint used to obtain the width of the middle cerebellar peduncle is located at this coordinate. Among them, the coordinates in the second target magnetic resonance image correspond to the coordinates in the target endpoint probability distribution image, that is, the coordinates in the target endpoint probability distribution image can be reflected in the second target magnetic resonance image. Referring to Figure 9 , there are two maximum probabilities in the figure. One is the distribution probability corresponding to the point m on the X-axis, and the other is the distribution probability corresponding to the point n on the X-axis.

[0141] Step S722: Use the first original coordinate corresponding to the first maximum probability as the target endpoint coordinate.

[0142] Specifically, referring to Figure 9 , the distribution probabilities corresponding to point m and point n are the first maximum probabilities. Therefore, point m and point n are the target endpoint coordinates.

[0143] Suppose there are 20 second target magnetic resonance images input into the target endpoint detection model, including 10 images of the left middle cerebellar peduncle area and 10 images of the right middle cerebellar peduncle area. Then the target endpoint detection model correspondingly outputs 20 target endpoint probability distribution images. Taking the 10 target endpoint probability distribution images corresponding to the left middle cerebellar peduncle as an example, each of these 10 target endpoint probability distribution images has two maximum probabilities similar to those in Figure 9 . Calculate the mean of the two maximum probabilities in each image, compare the means of each image, and select the image with the largest mean as the correct image. Confirm the above two maximum probabilities from this correct image, and use the two first original coordinates corresponding to the two maximum probabilities as the target endpoint coordinates of the left middle cerebellar peduncle. Similarly, the target endpoint coordinates of the right middle cerebellar peduncle can also be obtained in the above manner. Finally, calculate the width of the left middle cerebellar peduncle according to the target endpoint coordinates of the left middle cerebellar peduncle, calculate the width of the right middle cerebellar peduncle according to the target endpoint coordinates of the right middle cerebellar peduncle, and use the mean of the width of the left middle cerebellar peduncle and the width of the right middle cerebellar peduncle as the middle cerebellar peduncle data.

[0144] As shown in Figure 10As shown, in some embodiments of the present application, before step S710, the image processing method further includes training a target end point detection model, specifically including but not limited to steps S711 to S715. The following provides a detailed introduction to these five steps.

[0145] Step S711: Obtain a sample image of the middle cerebellar peduncle and the sample end point coordinates of the sample image of the middle cerebellar peduncle.

[0146] Specifically, the sample image of the middle cerebellar peduncle is a sectional image of the middle cerebellar peduncle on the midsagittal plane. The sample end points are obtained by annotation and are used to calculate the width of the middle cerebellar peduncle.

[0147] Step S712: Input the sample image of the middle cerebellar peduncle into a preset original end point detection model for detection processing to obtain an original end point probability distribution image; wherein, the original end point probability distribution image includes second original coordinates and the second distribution probability of the second original coordinates.

[0148] Specifically, the preset original end point detection model can be an HRNet network model. The original end point detection model detects the end point distribution probability based on the input sample image of the middle cerebellar peduncle and outputs an original end point probability distribution map. The original end point probability distribution image can be an image as shown in Figure 11 where the X-axis is the second original coordinate and the Y-axis is the second distribution probability of the second original coordinate.

[0149] Step S713: Compare the second distribution probabilities to obtain the maximum second probability.

[0150] Step S714: Use the second original coordinate corresponding to the maximum second probability as the original end point coordinate.

[0151] Specifically, referring to Figure 11 , there are two maximum probabilities in the figure. One is the distribution probability corresponding to the point m1 on the X-axis, and the other is the distribution probability corresponding to the point n1 on the X-axis. Therefore, the points m1 and n1 can be selected as the original end point coordinates.

[0152] Step S715: Adjust the parameters of the original end point detection model according to the original end point coordinates and the sample end point coordinates to obtain a target end point detection model.

[0153] Specifically, compare the original end point coordinates with the sample end point coordinates to obtain the difference between the original end point coordinates and the sample end point coordinates. Subsequently, adjust the parameters of the original end point detection model according to this difference to obtain a target end point detection model. Here, reference can be made to Figure 9 、 Figure 11 , Figure 9 is the target end point probability distribution image output by the target boundary point detection model, Figure 11It is the original endpoint probability distribution image output by the original endpoint detection model. It can be seen that the endpoint distribution probability curve in the target endpoint probability distribution image is more convergent than the curve in the original endpoint probability distribution image, that is, the endpoint distribution probability in the target endpoint probability distribution image is more accurate than that in the original endpoint probability distribution image. Therefore, the target endpoint detection model trained in this embodiment can obtain a more accurate target endpoint probability distribution image.

[0154] As Figure 12 shown, in some embodiments of the present application, step S800 includes but is not limited to step S810 and step S830. The following will introduce these three steps in detail.

[0155] Step S810: Obtain the centroid coordinates and the bisector line according to the second magnetic resonance image; wherein, the centroid coordinates are used to represent the coordinates of the centroid of the superior cerebellar peduncle.

[0156] Specifically, referring to Figure 4 , this image is the second magnetic resonance image, which includes the left superior cerebellar peduncle region 400 and the right superior cerebellar peduncle region 500. Taking the left superior cerebellar peduncle region 400 as an example, the centroid is point z, which represents the mass center point of the superior cerebellar peduncle. Along the direction with the longest length of the left superior cerebellar peduncle through the centroid z, a straight line D that bisects the superior cerebellar peduncle is drawn, and this straight line D is used as the bisector line.

[0157] Step S820: Obtain the width line according to the centroid coordinates and the bisector line.

[0158] Specifically, referring to Figure 4 , a perpendicular line segment Lscp to the bisector line D is drawn through the centroid z. Both ends of this line segment Lscp are connected to the edge of the left superior cerebellar peduncle region 400, and the line segment Lscp is used as the width line of the left superior cerebellar peduncle.

[0159] Step S830: Obtain the superior cerebellar peduncle data according to the width line and the second magnetic resonance image.

[0160] Specifically, referring to Figure 4 , finally, the line length of the width line Lscp is calculated to obtain the width of the left superior cerebellar peduncle. Similarly, the width of the right superior cerebellar peduncle can also be obtained in the above manner. Finally, the average value of the width of the left superior cerebellar peduncle and the width of the right superior cerebellar peduncle is obtained, and this average value is used as the superior cerebellar peduncle data.

[0161] As Figure 13 shown, the embodiment of the present application also provides an image processing system, including:

[0162] An image acquisition module 610 is configured to acquire a first magnetic resonance image and a second magnetic resonance image; wherein, the first magnetic resonance image is used to represent a sagittal plane image of the brain, the second magnetic resonance image is used to represent a coronal plane image of the brain, the first magnetic resonance image includes: a first target magnetic resonance image and a second target magnetic resonance image, the first target magnetic resonance image is an image including the brainstem region, the second target magnetic resonance image is an image including the middle cerebellar peduncle region, and the second magnetic resonance image is an image including the superior cerebellar peduncle region;

[0163] A brainstem image processing module 620 is configured to input the first target magnetic resonance image into a preset brainstem generation model for modeling processing to obtain a three-dimensional brainstem image; perform cutting processing on the three-dimensional brainstem image to obtain a brainstem cut image and a first coordinate; wherein, the first coordinate is used to represent the coordinate of the bottom end of the quadrigeminal region in the brainstem cut image; obtain the area of the brainstem cut image, and use the brainstem cut image corresponding to the minimum value of the area as the target brainstem image; wherein, use the first coordinate of the target brainstem image as the target point coordinate;

[0164] An area acquisition module 630 is configured to input the target brainstem image into a preset target boundary point detection model for detection processing to obtain boundary point coordinates; obtain the midbrain area according to the boundary point coordinates, the target point coordinates, and the target brainstem image, and obtain the pons area according to the boundary point coordinates, the target point coordinates, and the target brainstem image;

[0165] A middle cerebellar peduncle data acquisition module 640 is configured to input the second target magnetic resonance image into a preset target end point detection model for detection processing to obtain target end point coordinates, and obtain middle cerebellar peduncle data according to the target end point coordinates;

[0166] A superior cerebellar peduncle data acquisition module 650 is configured to obtain superior cerebellar peduncle data according to the second magnetic resonance image;

[0167] A target data calculation module 660 is configured to obtain target data according to the midbrain area, the pons area, the middle cerebellar peduncle data, and the superior cerebellar peduncle data.

[0168] It can be seen that the content in the above embodiments of the image processing method is applicable to the embodiments of this image processing system. The functions specifically implemented by the embodiments of this image processing system are the same as those of the above embodiments of the image processing method, and the beneficial effects achieved are also the same as those of the above embodiments of the image processing method.

[0169] An embodiment of this application further provides an electronic device, including: at least one memory; at least one processor; at least one computer program; the computer program is stored in the memory, and the processor executes at least one computer program to implement the image processing method described in any of the above embodiments.

[0170] It can be seen that the content in the above embodiments of the image processing method is applicable to the embodiments of this electronic device. The functions specifically implemented by the embodiments of this electronic device are the same as those of the above embodiments of the image processing method, and the beneficial effects achieved are also the same as those of the above embodiments of the image processing method.

[0171] Next, in conjunction with Figure 14 this, a detailed introduction to the electronic device according to the embodiments of the present application will be given.

[0172] As Figure 14 , Figure 14 Figure 7 schematically shows the hardware structure of an electronic device according to another embodiment. The electronic device includes:

[0173] A processor 710, which can be implemented in a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, etc., and is used to execute relevant programs to implement the technical solutions provided by the embodiments of the present disclosure;

[0174] A memory 720, which can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM), etc. The memory 720 can store an operating system and other application programs. When implementing the technical solutions provided by the embodiments of this specification through software or firmware, the relevant program codes are stored in the memory 720 and are called by the processor 710 to execute the image processing method of the embodiments of the present disclosure;

[0175] An input / output interface 730, which is used to implement information input and output;

[0176] A communication interface 740, which is used to implement communication and interaction between this device and other devices, and can communicate through a wired method (such as USB, network cable, etc.) or a wireless method (such as mobile network, WIFI, Bluetooth, etc.);

[0177] A bus 750, which transmits information between various components of the device (such as the processor 710, the memory 720, the input / output interface 730, and the communication interface 740);

[0178] Among them, the processor 710, the memory 720, the input / output interface 730, and the communication interface 740 are communicatively connected to each other inside the device through the bus 750.

[0179] The embodiments of the present application also provide a computer-readable storage medium storing computer-executable instructions for causing a computer to execute the image processing method described in any of the above embodiments.

[0180] It can be seen that the content in the above embodiments of the image processing method is applicable to the embodiments of this computer-readable storage medium. The functions specifically implemented by the embodiments of this computer-readable storage medium are the same as those of the above embodiments of the image processing method, and the beneficial effects achieved are also the same as those of the above embodiments of the image processing method.

[0181] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0182] Those of ordinary skill in the art can understand that all or some of the steps in the methods disclosed above, and the functional modules / units in systems and devices can be implemented as software, firmware, hardware, and their appropriate combinations.

[0183] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above drawings are used to distinguish similar objects and do not necessarily describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments of the present application described herein can be implemented in an order different from those illustrated or described herein. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units does not necessarily limit to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these process, method, product, or device.

[0184] It should be understood that in this application, "at least one (item)" means one or more, and "a plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that there can be three relationships. For example, "A and / or B" can mean: only A exists, only B exists, and both A and B exist at the same time. Among them, A and B can be singular or plural. The character " / " generally indicates that the associated objects before and after are in an "or" relationship. "At least one (one) of the following" or similar expressions refer to any combination of these items, including any combination of single items (ones) or plural items (ones). For example, at least one (one) of a, b, or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0185] In several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only illustrative. For example, the division of units is only a logical function division. In actual implementation, there can be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection to each other can be through some interfaces. The indirect coupling or communication connection of devices or units can be in electrical, mechanical or other forms.

[0186] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place, or they can be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0187] In addition, each functional unit in various embodiments of this application can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.

[0188] When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions for causing an electronic device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of various embodiments of this application. The aforementioned storage medium includes: various media that can store programs, such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs.

[0189] The preferred embodiments of the embodiments of the present disclosure have been described above with reference to the accompanying drawings. This does not limit the scope of the rights of the embodiments of the present disclosure. Any modifications, equivalent replacements, and improvements made by those skilled in the art without departing from the scope and essence of the embodiments of the present disclosure shall be within the scope of the rights of the embodiments of the present disclosure.

Claims

1. An image processing method, characterized in that, Including: Obtaining a first magnetic resonance image and a second magnetic resonance image; wherein, the first magnetic resonance image is used to represent a sagittal plane image of the brain, the second magnetic resonance image is used to represent a coronal plane image of the brain, the first magnetic resonance image includes: a first target magnetic resonance image and a second target magnetic resonance image, the first target magnetic resonance image is an image containing the brainstem region, the second target magnetic resonance image is an image containing the middle cerebellar peduncle region, and the second magnetic resonance image is an image containing the superior cerebellar peduncle region; Inputting the first target magnetic resonance image into a preset brainstem generation model for modeling processing to obtain a three-dimensional brainstem image; Performing cutting processing on the three-dimensional brainstem image to obtain a brainstem cut image and a first coordinate; wherein, the first coordinate is used to represent the coordinate of the bottom end of the quadrigeminal region in the brainstem cut image; Obtaining the area of the brainstem cut image, and using the brainstem cut image corresponding to the minimum value of the area as the target brainstem image; wherein, using the first coordinate of the target brainstem image as the target point coordinate; Inputting the target brainstem image into a preset target boundary point detection model for detection processing to obtain boundary point coordinates; Obtaining the midbrain area according to the boundary point coordinates, the target point coordinates, and the target brainstem image, and obtaining the pons area according to the boundary point coordinates, the target point coordinates, and the target brainstem image; Inputting the second target magnetic resonance image into a preset target endpoint detection model for detection processing to obtain target endpoint coordinates, and obtaining middle cerebellar peduncle data according to the target endpoint coordinates; Obtaining superior cerebellar peduncle data according to the second magnetic resonance image; Obtaining target data according to the midbrain area, the pons area, the middle cerebellar peduncle data, and the superior cerebellar peduncle data; The boundary point coordinates include: a first sub-boundary point coordinate and a second sub-boundary point coordinate; The obtaining the midbrain area according to the boundary point coordinates, the target point coordinates, and the target brainstem image, and obtaining the pons area according to the boundary point coordinates, the target point coordinates, and the target brainstem image includes: Obtaining a first boundary line according to the first sub-boundary point coordinate and the target point coordinate; Obtaining a midbrain region according to the first boundary line and the target brainstem image, and obtaining the midbrain area according to the midbrain region; Obtaining a second boundary line according to the second sub-boundary point coordinate and the first boundary line; wherein, the second boundary line is parallel to the first boundary line; Obtaining a pons region according to the first boundary line, the second boundary line, and the target brainstem image, and obtaining the pons area according to the pons region.

2. The image processing method according to claim 1, wherein Before the inputting the target brainstem image into a preset target boundary point detection model for detection processing to obtain boundary point coordinates, the method further includes training the target boundary point detection model, specifically including: Obtaining a sample brainstem image and sample boundary point coordinates of the sample brainstem image; Inputting the sample brainstem image into a preset original boundary point detection model for detection processing to obtain original boundary point coordinates; Adjust the parameters of the original boundary point detection model according to the original boundary point coordinates and the sample boundary point coordinates to obtain a target boundary point detection model.

3. The image processing method according to claim 1, characterized in that The step of inputting the second target magnetic resonance image into a preset target endpoint detection model for detection processing to obtain target endpoint coordinates includes: Input the second target magnetic resonance image into the target endpoint detection model for detection processing to obtain a target endpoint probability distribution image; Obtain the target endpoint coordinates according to the target endpoint probability distribution image.

4. The image processing method according to claim 3, characterized in that, The target endpoint probability distribution image includes first original coordinates and a first distribution probability of the first original coordinates; The step of obtaining the target endpoint coordinates according to the target endpoint probability distribution image includes: Compare the first distribution probabilities to obtain a maximum first probability; Use the first original coordinates corresponding to the maximum first probability as the target endpoint coordinates.

5. The image processing method according to claim 4, characterized in that Before inputting the second target magnetic resonance image into the target endpoint detection model for detection processing to obtain a target endpoint probability distribution image, the method further includes: training the target endpoint detection model, specifically including: Obtain a sample middle cerebellar peduncle image and sample endpoint coordinates of the sample middle cerebellar peduncle image; Input the sample middle cerebellar peduncle image into a preset original endpoint detection model for detection processing to obtain an original endpoint probability distribution image; wherein, the original endpoint probability distribution image includes second original coordinates and a second distribution probability of the second original coordinates; Compare the second distribution probabilities to obtain a maximum second probability; Use the second original coordinates corresponding to the maximum second probability as the original endpoint coordinates; Adjust the parameters of the original endpoint detection model according to the original endpoint coordinates and the sample endpoint coordinates to obtain a target endpoint detection model.

6. The image processing method according to any one of claims 1 to 5, characterized in that, The step of obtaining superior cerebellar peduncle data according to the second magnetic resonance image includes: Obtain centroid coordinates and a bisector according to the second magnetic resonance image; wherein, the centroid coordinates are used to represent the coordinates of the centroid of the superior cerebellar peduncle; Obtain a width line according to the centroid coordinates and the bisector; Obtain superior cerebellar peduncle data according to the width line and the second magnetic resonance image.

7. An image processing system, characterized in that, including: An image acquisition module, which is used to acquire a first magnetic resonance image and a second magnetic resonance image; wherein, the first magnetic resonance image is used to represent a sagittal plane image of the brain, the second magnetic resonance image is used to represent a coronal plane image of the brain, the first magnetic resonance image includes: a first target magnetic resonance image and a second target magnetic resonance image, the first target magnetic resonance image is an image containing the brainstem region, the second target magnetic resonance image is an image containing the middle cerebellar peduncle region, and the second magnetic resonance image is an image containing the superior cerebellar peduncle region; Brainstem image processing module, which is configured to input the first target magnetic resonance image into a preset brainstem generation model for modeling processing to obtain a three-dimensional brainstem image; perform cutting processing on the three-dimensional brainstem image to obtain a brainstem cut image and a first coordinate; wherein, the first coordinate is used to represent the coordinate of the bottom end of the quadrigeminal region in the brainstem cut image; obtain the area of the brainstem cut image, and use the brainstem cut image corresponding to the minimum value of the area as the target brainstem image; wherein, use the first coordinate of the target brainstem image as the target point coordinate; Area acquisition module, which is configured to input the target brainstem image into a preset target boundary point detection model for detection processing to obtain boundary point coordinates; obtain the midbrain area according to the boundary point coordinates, the target point coordinate, and the target brainstem image, and obtain the pons area according to the boundary point coordinates, the target point coordinate, and the target brainstem image; Middle cerebellar peduncle data acquisition module, which is configured to input the second target magnetic resonance image into a preset target endpoint detection model for detection processing to obtain target endpoint coordinates, and obtain middle cerebellar peduncle data according to the target endpoint coordinates; Superior cerebellar peduncle data acquisition module, which is configured to obtain superior cerebellar peduncle data according to the second magnetic resonance image; Target data calculation module, which is configured to obtain target data according to the midbrain area, the pons area, the middle cerebellar peduncle data, and the superior cerebellar peduncle data; The boundary point coordinates include: first sub-boundary point coordinates and second sub-boundary point coordinates; The step of obtaining the midbrain area according to the boundary point coordinates, the target point coordinate, and the target brainstem image, and obtaining the pons area according to the boundary point coordinates, the target point coordinate, and the target brainstem image includes: Obtain a first boundary line according to the first sub-boundary point coordinates and the target point coordinate; Obtain a midbrain region according to the first boundary line and the target brainstem image, and obtain the midbrain area according to the midbrain region; Obtain a second boundary line according to the second sub-boundary point coordinates and the first boundary line; wherein, the second boundary line is parallel to the first boundary line; Obtain a pons region according to the first boundary line, the second boundary line, and the target brainstem image, and obtain the pons area according to the pons region.

8. An electronic device, characterized in that, Comprising: At least one memory; At least one processor; At least one computer program; The computer program is stored in the memory, and the processor executes the at least one computer program to implement the image processing method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to cause a computer to execute the image processing method according to any one of claims 1 to 6.

Images