Brain image processing method, computer device and storage medium

Abstract

This application relates to a brain image processing method, computer device, and storage medium. The method includes: inputting a brain image to be analyzed into a preset first segmentation model to obtain a segmentation result of a region of interest (ROI) in the brain image; the ROI is a brain region within a preset range surrounding the trigeminal nerve in the brain image; the first segmentation model is obtained by training a preset first initial segmentation model based on a sample brain image and a gold standard for brain region segmentation of the sample brain image; inputting the segmentation result of the ROI and the brain image into a preset second segmentation model to obtain a segmentation result of the trigeminal nerve; the second segmentation model is obtained by training a preset second initial segmentation model based on a sample brain image, a gold standard for brain region segmentation of the sample brain image, and a gold standard for trigeminal nerve segmentation of the sample brain image. This method can automatically classify the pain level of trigeminal neuralgia based on medical images.

Description

Technical Field

[0001] This application relates to the field of image processing technology, and in particular to a brain image processing method, computer device, and storage medium. Background Technology

[0002] Trigeminal neuralgia is a type of neuropathic facial pain, typically characterized by severe, stabbing pain on one side of the face, with brief, sudden onset and recurring episodes. Therefore, timely and appropriate treatment based on the pain level of trigeminal neuralgia can effectively prevent the pain from worsening. Thus, accurate pain level classification for trigeminal neuralgia is particularly important.

[0003] Traditional techniques primarily rely on patients' subjective descriptions of pain to evaluate and determine their pain levels. However, traditional methods have the limitation of not being able to automatically classify pain levels based on medical images. Summary of the Invention

[0004] Therefore, it is necessary to provide a brain image processing method, computer device, and storage medium that can automatically classify the pain level of trigeminal neuralgia based on medical images to address the above-mentioned technical problems.

[0005] In a first aspect, this application provides a brain image processing method, the method comprising:

[0006] The brain image to be analyzed is input into a preset first segmentation model to obtain the segmentation result of the region of interest in the brain image; the region of interest is the brain region within a preset range around the trigeminal nerve in the brain image; the first segmentation model is obtained by training a preset first initial segmentation model based on the sample brain image and the gold standard for brain region segmentation of the sample brain image.

[0007] The segmentation results of the region of interest and the brain image are input into a preset second segmentation model to obtain the segmentation results of the trigeminal nerve; the second segmentation model is obtained by training a preset second initial segmentation model based on the sample brain image, the gold standard for brain region segmentation of the sample brain image, and the gold standard for trigeminal nerve segmentation of the sample brain image.

[0008] In one embodiment, the method further includes:

[0009] The brain image, the segmentation result, and the clinical representation information corresponding to the brain image are input into a preset evaluation network to obtain the evaluation result corresponding to the brain image; the evaluation result is used to characterize the feature information of the trigeminal nerve.

[0010] In one embodiment, the evaluation result includes a positive / negative classification result and the type of the positive category; the step of inputting the brain image, the segmentation result, and the clinical representation information corresponding to the brain image into a preset evaluation network to obtain the evaluation result corresponding to the brain image includes:

[0011] The brain image, the segmentation result, and the corresponding clinical representation information of the brain image are input into the first classification network in the evaluation network to determine the first classification result of the brain image; the first classification result is used to characterize whether the brain image is a positive or negative category.

[0012] When the first classification result is a positive category, the brain image, the segmentation result, and the clinical representation information corresponding to the brain image are input into the second classification network in the evaluation network to determine the second classification result of the brain image; the second classification result is used to characterize the type to which the brain image belongs, and the type to which the brain image belongs includes negative and the type of positive category to which the brain image belongs when it is a positive category.

[0013] In one embodiment, the segmentation result of the trigeminal nerve is a mask image of the trigeminal nerve; the method further includes:

[0014] Based on the brain image and the mask image, a segmentation map of the trigeminal nerve is obtained.

[0015] In one embodiment, the step of inputting the brain image to be analyzed into a preset first segmentation model to obtain a first mask map of the region of interest in the brain image includes:

[0016] The image type of the brain image is determined by a preset discriminator; the image type is either a 2D image or a 3D image.

[0017] If the brain image is a 2D image, then the brain image is input into the first segmentation sub-model to obtain the first mask image; or,

[0018] If the brain image is a 3D image, then the brain image is input into the second segmentation sub-model to obtain the first mask image.

[0019] In one embodiment, the method further includes:

[0020] The gold standard for brain region segmentation of the sample brain image and the sample brain image are input into the second initial segmentation model to obtain the sample segmentation result of the trigeminal nerve of the sample brain image;

[0021] Based on the sample segmentation results and the gold standard for trigeminal nerve segmentation in the sample brain image, the value of the first loss function of the second initial segmentation model is obtained;

[0022] The second initial segmentation model is trained based on the value of the first loss function to obtain the second segmentation model.

[0023] In one embodiment, the method further includes:

[0024] Obtain the gold standard evaluation results corresponding to the brain images of the samples;

[0025] The sample brain image, the gold standard for segmentation of the trigeminal nerve in the sample brain image, and the sample clinical characterization information corresponding to the sample brain image are input into a preset initial evaluation network to obtain the sample evaluation result corresponding to the sample brain image.

[0026] Based on the sample evaluation results and the gold standard evaluation results, the value of the second loss function of the initial evaluation network is obtained;

[0027] The initial evaluation network is trained based on the value of the second loss function to obtain the evaluation network.

[0028] In one embodiment, the first loss function includes the Dice Loss function and the CL loss function, and the second loss function includes the Focal loss function and the CAM loss function.

[0029] Secondly, this application also provides a brain image processing apparatus, the apparatus comprising:

[0030] The first processing module is used to input the brain image to be analyzed into a preset first segmentation model to obtain the segmentation result of the region of interest in the brain image; the region of interest is the brain region within a preset range around the trigeminal nerve in the brain image; the first segmentation model is obtained by training the preset first initial segmentation model according to the sample brain image and the gold standard for brain region segmentation of the sample brain image.

[0031] The second processing module is used to input the segmentation result of the region of interest and the brain image into a preset second segmentation model to obtain the segmentation result of the trigeminal nerve; the second segmentation model is obtained by training a preset second initial segmentation model based on the sample brain image, the gold standard for brain region segmentation of the sample brain image, and the gold standard for trigeminal nerve segmentation of the sample brain image.

[0032] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to perform the following steps:

[0033] The brain image to be analyzed is input into a preset first segmentation model to obtain the segmentation result of the region of interest in the brain image; the region of interest is the brain region within a preset range around the trigeminal nerve in the brain image; the first segmentation model is obtained by training a preset first initial segmentation model based on the sample brain image and the gold standard for brain region segmentation of the sample brain image.

[0034] The segmentation results of the region of interest and the brain image are input into a preset second segmentation model to obtain the segmentation results of the trigeminal nerve; the second segmentation model is obtained by training a preset second initial segmentation model based on the sample brain image, the gold standard for brain region segmentation of the sample brain image, and the gold standard for trigeminal nerve segmentation of the sample brain image.

[0035] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, performs the following steps:

[0036] The brain image to be analyzed is input into a preset first segmentation model to obtain the segmentation result of the region of interest in the brain image; the region of interest is the brain region within a preset range around the trigeminal nerve in the brain image; the first segmentation model is obtained by training a preset first initial segmentation model based on the sample brain image and the gold standard for brain region segmentation of the sample brain image.

[0037] The segmentation results of the region of interest and the brain image are input into a preset second segmentation model to obtain the segmentation results of the trigeminal nerve; the second segmentation model is obtained by training a preset second initial segmentation model based on the sample brain image, the gold standard for brain region segmentation of the sample brain image, and the gold standard for trigeminal nerve segmentation of the sample brain image.

[0038] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, performs the following steps:

[0039] The brain image to be analyzed is input into a preset first segmentation model to obtain the segmentation result of the region of interest in the brain image; the region of interest is the brain region within a preset range around the trigeminal nerve in the brain image; the first segmentation model is obtained by training a preset first initial segmentation model based on the sample brain image and the gold standard for brain region segmentation of the sample brain image.

[0040] The segmentation results of the region of interest and the brain image are input into a preset second segmentation model to obtain the segmentation results of the trigeminal nerve; the second segmentation model is obtained by training a preset second initial segmentation model based on the sample brain image, the gold standard for brain region segmentation of the sample brain image, and the gold standard for trigeminal nerve segmentation of the sample brain image.

[0041] The aforementioned brain image processing method, computer equipment, and storage medium, because the preset first segmentation model is obtained by training a preset first initial segmentation model based on the sample brain image and the gold standard for brain region segmentation of the sample brain image, can accurately obtain the segmentation result of the region of interest within a preset range around the trigeminal nerve in the brain image by inputting the brain image to be analyzed into the trained first segmentation model. This allows the segmentation result of the region of interest and the brain image to be input into a preset second segmentation model. Since the second segmentation model is obtained by training a preset second initial segmentation model based on the sample brain image, the gold standard for brain region segmentation of the sample brain image, and the gold standard for trigeminal nerve segmentation of the sample brain image, from... The second segmentation model can accurately segment the trigeminal nerve in the brain image, obtaining the segmentation result of the trigeminal nerve. Then, the pain level of trigeminal neuralgia can be classified using the obtained trigeminal nerve segmentation result. Since this process involves inputting the brain image to be analyzed into the first segmentation model to obtain the segmentation result of the region of interest, and inputting the segmentation result of the region of interest and the brain image into the second segmentation model to obtain the trigeminal nerve segmentation result, human intervention is reduced. The final trigeminal nerve segmentation result can automatically classify the pain level of trigeminal neuralgia, solving the problem that traditional technologies cannot automatically classify pain levels based on medical images. Attached Figure Description

[0042] Figure 1 This is a diagram illustrating the application environment of a brain image processing method in one embodiment;

[0043] Figure 2 This is a flowchart illustrating a brain image processing method in one embodiment;

[0044] Figure 3 This is a flowchart illustrating a brain image processing method in another embodiment;

[0045] Figure 4 This is a flowchart illustrating a brain image processing method in another embodiment;

[0046] Figure 5 This is a flowchart illustrating a brain image processing method in another embodiment;

[0047] Figure 6 This is a flowchart illustrating a brain image processing method in another embodiment;

[0048] Figure 7 This is a flowchart illustrating a brain image processing method in another embodiment;

[0049] Figure 8 This is a flowchart illustrating a brain image processing method in another embodiment;

[0050] Figure 9 This is a flowchart illustrating a brain image processing method in another embodiment;

[0051] Figure 10 This is a structural block diagram of a brain image processing device in one embodiment. Detailed Implementation

[0052] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0053] Trigeminal neuralgia is a type of neuropathic facial pain, typically characterized by severe, stabbing pain on one side of the face, with brief, sudden onset and recurrent episodes. Current research classifies trigeminal neuralgia into three categories based on etiology: primary trigeminal neuralgia, classic trigeminal neuralgia, and secondary trigeminal neuralgia. Primary trigeminal neuralgia has no obvious cause, classic trigeminal neuralgia is caused by vascular compression of the trigeminal nerve root, and secondary trigeminal neuralgia is secondary to neurological diseases such as tumor compression or multiple sclerosis.

[0054] Current diagnostic criteria for trigeminal neuralgia primarily rely on patients' descriptions of their pain attacks. While the characteristics of trigeminal neuralgia are clearly defined, inconsistencies in descriptive terminology hinder communication between patients, physicians, and researchers, further impacting clinical assessment. On the other hand, some researchers have studied MRI brain images of trigeminal neuralgia patients and healthy individuals. These studies involve professional radiologists delineating the trigeminal nerve region on MRI images to identify regions of interest (ROIs). This is then combined with clinical information and radiomics features extracted from the ROIs to infer the pathogenesis and etiology. However, annotating the trigeminal nerve region on MRI images is mainly done by experienced physicians, which is slow and labor-intensive. Furthermore, individual differences in patient images and the subjectivity of the annotators further increase the difficulty and unreliability of manual annotation. While MR imaging is widely considered the preferred method for examining the trigeminal nerve and its root, showing whether the nerve root is in contact with neurovascular structures, conventional imaging is often insufficient to identify the cause of trigeminal neuralgia in patients without trigeminal symptoms. These studies have not yet been able to link imaging features to the actual pain experienced by trigeminal neuralgia patients, nor can they automatically classify the severity of trigeminal neuralgia. Therefore, this application uses a discriminator to filter input data and selects a corresponding brain region segmentation network as the localization network for trigeminal nerve segmentation. This localization network provides candidate bounding boxes for trigeminal nerve segmentation and segmentation labels for adjacent brain regions. Based on the localization network, a segmentation mask for the trigeminal nerve's adjacent regions is obtained. This brain region segmentation mask and the original input image are then used as dual-channel inputs to a second-stage V-shaped convolutional neural network for co-localization, achieving accurate trigeminal nerve segmentation. Based on the trigeminal nerve segmentation, machine learning or deep learning methods are used for trigeminal neuralgia grading assessment, enabling automatic classification of trigeminal neuralgia pain levels based on medical images.

[0055] The brain image processing method provided in this application embodiment can be applied to, for example, Figure 1The computer device shown includes a processor and a memory connected via a system bus. The memory stores a computer program, and the processor executes the computer program to perform the steps described in the method embodiments below. Optionally, the computer device may further include a network interface, a display screen, and an input device. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with external terminals via a network connection. Optionally, the computer device may be a server, a personal computer, a personal digital assistant, or other terminal devices, such as tablet computers, mobile phones, etc., or it may be a cloud or remote server. This application embodiment does not limit the specific form of the computer device.

[0056] In one embodiment, such as Figure 2 As shown, a brain image processing method is provided, which can be applied to... Figure 1 Taking a computer device as an example, the explanation includes the following steps:

[0057] S201, Input the brain image to be analyzed into the preset first segmentation model to obtain the segmentation result of the region of interest in the brain image; the region of interest is the brain region within a preset range around the trigeminal nerve in the brain image; the first segmentation model is obtained by training the preset first initial segmentation model based on the sample brain image and the gold standard of brain region segmentation of the sample brain image.

[0058] Optionally, in this embodiment, the segmentation result of the region of interest (ROI) in the brain image can be either a mask map of the ROI or a segmentation map of the ROI. The mask map is a binary image composed of 0s and 1s. When the mask is applied to the image, regions with a value of 1 are processed, while regions with a value of 0 are not included in the calculation. Typically, the image mask can be defined by specifying at least one of the following: a data value, a data range, a finite or infinite value, and the ROI. It is understood that image segmentation divides an image into several specific regions with unique properties. In this embodiment, the aforementioned ROI segmentation map refers to a local image of the ROI segmented from the brain image. It is understood that multiplying the aforementioned ROI mask map and the aforementioned brain image yields the local image of the ROI.

[0059] Optionally, the first segmentation model in this embodiment can be any one of the U-Net model, 3D U-Net model, V-Net model, and ResNet model. Optionally, the brain image to be analyzed in this embodiment can be a computed tomography (CT) image, a magnetic resonance imaging (MRI) image, etc., and further, the brain image can be a 3D brain image or a 2D brain image. For example, the region of interest in this embodiment can be the pons adjacent to the trigeminal nerve, the periorbital region, and the angle of the mandible. Optionally, in this embodiment, the sample brain image can be input into a preset first initial segmentation model to obtain the brain region sample segmentation result of the sample brain image. Based on the brain region sample segmentation result of the sample brain image and the gold standard for brain region segmentation of the sample brain image, the value of the loss function of the first initial segmentation model is obtained. The first initial segmentation model is trained using the value of the loss function of the first initial segmentation model to obtain the above-mentioned first segmentation model. Optionally, the above-mentioned loss function can be the Dice Loss function, etc.

[0060] It is understood that the segmentation results of the region of interest in the brain image in this embodiment may include information such as the location and shape of the region of interest.

[0061] S202, input the segmentation results of the region of interest and the brain image into the preset second segmentation model to obtain the segmentation results of the trigeminal nerve; the second segmentation model is obtained by training the preset second initial segmentation model based on the sample brain image, the gold standard for brain region segmentation of the sample brain image, and the gold standard for trigeminal nerve segmentation of the sample brain image.

[0062] Optionally, in this embodiment, the second segmentation model can be, but is not limited to, the U-Net model, the 3D U-Net model, the V-Net model, etc., and can have various variations.

[0063] In this embodiment, exemplarily, such as Figure 3As shown, the computer device can use the segmentation result of the region of interest (ROI) and the brain image to be analyzed as dual-channel inputs to a preset second segmentation model. The second segmentation model then obtains the segmentation result of the trigeminal nerve. Optionally, the segmentation result of the trigeminal nerve in this embodiment can be a mask image of the trigeminal nerve or a segmentation image of the trigeminal nerve. Optionally, in this embodiment, after inputting the segmentation result of the ROI and the brain image to be analyzed into the second segmentation model, the second segmentation model can determine the range of the ROI in the brain image based on the ROI segmentation result, and then segment the brain image within the range of the ROI. The resulting segmented image is determined as the segmentation result of the trigeminal nerve. In other words, the segmentation result of the ROI in the brain image can limit the segmentation to the region corresponding to the ROI in the original brain image. The input of the ROI segmentation result provides proximity information of the trigeminal nerve in terms of location, enhancing the segmentation accuracy of the second segmentation model. Optionally, in this embodiment, the above-mentioned sample brain image and the gold standard for brain region segmentation of the sample brain image can be input into a preset second initial segmentation model to obtain the sample segmentation result of the trigeminal nerve of the sample brain image. The second initial segmentation model is trained based on the sample segmentation result of the trigeminal nerve of the sample brain image and the gold standard for segmentation of the trigeminal nerve of the sample brain image to obtain the above-mentioned second segmentation model.

[0064] In the aforementioned brain image processing method, since the preset first segmentation model is obtained by training a preset first initial segmentation model based on the sample brain image and the gold standard for brain region segmentation of the sample brain image, by inputting the brain image to be analyzed into the trained first segmentation model, the segmentation result of the region of interest within a preset range around the trigeminal nerve in the brain image can be accurately obtained. Therefore, the segmentation result of the region of interest and the brain image can be input into a preset second segmentation model. Since the second segmentation model is obtained by training a preset second initial segmentation model based on the sample brain image, the gold standard for brain region segmentation of the sample brain image, and the gold standard for trigeminal nerve segmentation of the sample brain image, it can accurately obtain the segmentation result of the region of interest within a preset range around the trigeminal nerve in the brain image. The second segmentation model accurately segments the trigeminal nerve in the brain image, obtaining the segmentation result of the trigeminal nerve. This segmentation result can then be used to classify the pain level of trigeminal neuralgia. Since this process involves inputting the brain image to be analyzed into the first segmentation model to obtain the segmentation result of the region of interest, and then inputting the segmentation result of the region of interest and the brain image into the second segmentation model to obtain the segmentation result of the trigeminal nerve, human intervention is reduced. The final trigeminal nerve segmentation result enables automatic classification of the pain level of trigeminal neuralgia, solving the problem in traditional techniques that cannot automatically classify pain levels based on medical images.

[0065] After obtaining the segmentation results of the trigeminal nerve, in some scenarios, an evaluation result of the brain image can be obtained based on the obtained segmentation results of the trigeminal nerve, the brain image to be analyzed, and the clinical representation information corresponding to the brain image. In one embodiment, the above method further includes: inputting the brain image, the segmentation results of the trigeminal nerve, and the clinical representation information corresponding to the brain image into a preset evaluation network to obtain the evaluation result corresponding to the brain image; the evaluation result is used to characterize the feature information of the trigeminal nerve.

[0066] The clinical representation information corresponding to the brain image can be the user's height, weight, age, etc. Optionally, the evaluation network in this embodiment can be any one of the following networks: U-Net network, 3D U-Net network, V-Net network, ResNet network, etc.

[0067] It is understood that the evaluation result corresponding to the brain image in this embodiment represents the characteristic information of the trigeminal nerve in the brain image. Through this characteristic information, the pain level of the trigeminal neuralgia of the user corresponding to the brain image can be obtained. For example, the evaluation result in this embodiment can represent that the user corresponding to the brain image is a user without trigeminal neuralgia, or it can represent that the pain level of the trigeminal neuralgia of the user corresponding to the brain image is level one, or it can represent that the pain level of the trigeminal neuralgia of the user corresponding to the brain image is level two.

[0068] It is understood that, in this embodiment, the three-channel input of the evaluation network can not only focus the network's attention on the trigeminal nerve region, but also focus on the peripheral region of the trigeminal nerve through the original brain image. At the same time, it can combine imaging information with clinical information to obtain the evaluation results corresponding to the brain image that characterizes the trigeminal nerve.

[0069] In this embodiment, by inputting brain images, trigeminal nerve segmentation results, and corresponding clinical representation information into a preset evaluation network, evaluation results representing the trigeminal nerve feature information can be obtained. This process avoids human intervention and can quickly and accurately obtain the evaluation results corresponding to the brain images to be analyzed, thus improving the efficiency and accuracy of obtaining the evaluation results corresponding to the brain images to be analyzed.

[0070] In the scenario described above, where the brain image is evaluated via an evaluation network, the evaluation result may include a positive / negative classification result and the type of the positive category. In one embodiment, such as... Figure 4 As shown, the above-mentioned "inputting brain images, segmentation results, and corresponding clinical representation information of brain images into a preset evaluation network to obtain evaluation results corresponding to brain images" includes:

[0071] S301, input the brain image, segmentation results and corresponding clinical representation information of the brain image into the first classification network in the evaluation network to determine the first classification result of the brain image; the first classification result is used to characterize the brain image as a positive or negative category.

[0072] Optionally, the first classification network in this embodiment can be any one of the following networks: U-Net, 3D U-Net, V-Net, ResNet, etc. It is understood that if the first classification result indicates that the brain image is a negative category, it means that the user corresponding to the brain image does not have trigeminal neuralgia; if the first classification result indicates that the brain image is a positive category, it means that the user corresponding to the brain image has trigeminal neuralgia, and the pain level of the user's trigeminal neuralgia can be further classified. Optionally, in this embodiment, the first classification network can analyze the brain image, the segmentation result of the trigeminal nerve, and the clinical representation information corresponding to the brain image through multiple convolutional layers and fixed-length fully connected layers, and perform binary classification of the user corresponding to the brain image into normal individuals and those with trigeminal neuralgia.

[0073] S302, when the first classification result is a positive category, the brain image, segmentation result and corresponding clinical representation information of the brain image are input into the second classification network in the evaluation network to determine the second classification result of the brain image; the second classification result is used to characterize the type to which the brain image belongs, and the type to which the brain image belongs includes negative and the type of positive category to which the brain image belongs when it is a positive category.

[0074] Optionally, the second classification network in this embodiment can be any one of the following networks: U-Net, 3D U-Net, V-Net, ResNet, etc. Optionally, in this embodiment, brain images, trigeminal nerve segmentation results, and corresponding clinical representation information of the brain images can be input into the second classification network. The second classification result, representing the type to which the brain image belongs, is determined through the convolutional layers and fully connected layers of the second classification network. Optionally, when a brain image is a positive category, the type of the positive category includes, but is not limited to, level one, level two, level three, and level four.

[0075] For example, in this embodiment, the process of obtaining the evaluation result corresponding to the brain image by evaluating the first classification network and the second classification network can be as follows: Figure 5 As shown, the trigeminal nerve grading assessment module is a two-level classification network with three input channels. The first channel is the brain image to be processed, the second channel is the segmentation result of the trigeminal nerve, and the third channel is the digitized clinical representation information of the brain image. This three-channel input not only allows the network to focus on the trigeminal nerve region but also allows attention to the surrounding areas of the trigeminal nerve through the original image. It also combines imaging information with clinical information. Specific details are as follows... Figure 5As shown, the trigeminal neuralgia grading assessment module consists of two cascaded sub-networks. The input to the first classification network is the brain image, the segmentation result of the trigeminal nerve, and the corresponding clinical representation information of the brain image. The first classification network performs binary classification on the brain image, the segmentation result of the trigeminal nerve, and the corresponding clinical representation information of the brain image through multiple convolutional layers and fixed-length fully connected layers, classifying the user corresponding to the brain image into normal individuals and trigeminal neuralgia patients. If the first classification network determines that the user corresponding to the brain image is a trigeminal neuralgia patient, the brain image, the segmentation result of the trigeminal nerve, and the corresponding clinical representation information of the brain image are input into the second classification network, which performs 4-level classification through convolutional layers and fully connected layers, classifying the trigeminal neuralgia of the user corresponding to the brain image into levels 1 to 4.

[0076] In this embodiment, by inputting the brain image to be analyzed, the segmentation result of the trigeminal nerve in the brain image, and the corresponding clinical representation information of the brain image into the first classification network of the evaluation network, it is possible to determine whether the brain image is a positive or negative category. Furthermore, when the brain image is a positive category, the brain image, the segmentation result of the trigeminal nerve in the brain image, and the corresponding clinical representation information of the brain image can be input into the second classification network of the evaluation network to determine the type of the brain image. No human intervention is required in this process, and the second classification result of the brain image can be automatically obtained through the evaluation network.

[0077] In some scenarios, it may also be necessary to use the segmentation map of the trigeminal nerve in the brain image mentioned above. In one embodiment, the segmentation result can be a mask map of the trigeminal nerve; the method further includes: obtaining the segmentation map of the trigeminal nerve based on the brain image and the mask map.

[0078] Optionally, in this embodiment, the computer device can multiply the brain image and the mask image of the trigeminal nerve in the brain image to obtain a segmentation map of the trigeminal nerve in the brain image. It can be understood that the mask image of the trigeminal nerve is a binary image composed of 0s and 1s, while the segmentation map of the trigeminal nerve is a segmentation map that separates the trigeminal nerve from the brain image, containing only the segmentation map of the trigeminal nerve.

[0079] In this embodiment, based on the brain image and the mask image of the trigeminal nerve in the brain image, the segmentation map of the trigeminal nerve can be obtained quickly and accurately, which improves the efficiency and accuracy of obtaining the segmentation map of the trigeminal nerve.

[0080] In some scenarios, the brain images to be analyzed may be 2D or 3D images. The computer device can first determine the image type of the brain image to be analyzed, and then input brain images of different image types into different segmentation models to segment the regions of interest in the brain images. In one embodiment, such as... Figure 6 As shown, the above S201 includes:

[0081] S401, determine the image type of the brain image through a preset discriminator; the image type is either a 2D image or a 3D image.

[0082] Optionally, in this embodiment, the number of image layers in the brain image can be determined by a preset discriminator, and the image type of the brain image can be determined based on the number of image layers. For example, if the discriminator determines that the number of layers in the brain image is one, the computer device can determine that the image type of the brain image is a 2D image; if the discriminator determines that the number of layers in the brain image is multiple, the computer device can determine that the image type of the brain image is a 3D image.

[0083] S402, if the brain image is a 2D image, then input the brain image into the first segmentation sub-model to obtain the first mask image.

[0084] For example, please see Figure 7 In this embodiment, if the brain image is a 2D image, the computer device can input the brain image into the first segmentation sub-model, perform segmentation processing on the 2D brain image, and obtain the first mask image.

[0085] S403, if the brain image is a 3D image, then input the brain image into the second segmentation sub-model to obtain the first mask image.

[0086] For example, please continue to see Figure 7 In this embodiment, if the brain image is a 3D image, the computer device can input the brain image into the second segmentation sub-model to segment the 3D image and obtain the first mask image. It is understood that the second segmentation sub-model in this embodiment is a model for processing 3D brain images; therefore, the number of convolutional layers in the second segmentation sub-model in this embodiment is greater than the number of convolutional layers in the first segmentation sub-model.

[0087] In this embodiment, a preset discriminator can determine the image type of the brain image to be analyzed, thereby allowing brain images of different image types to be input into the corresponding segmentation sub-model for accurate segmentation of the brain image and obtaining a mask map of the region of interest of the brain image, thus ensuring the accuracy of the obtained mask map of the region of interest of the brain image.

[0088] In the scenario described above, where the brain image to be analyzed is input into a first segmentation model to obtain a first mask of the region of interest in the brain image, and the first mask and the brain image are input into a second segmentation model to obtain a second mask of the trigeminal nerve, both the first and second segmentation models are pre-trained models. In one embodiment, such as... Figure 8 As shown, the above method also includes:

[0089] S501, input the gold standard for brain region segmentation of the sample brain image and the sample brain image into the second initial segmentation model to obtain the sample segmentation result of the trigeminal nerve in the sample brain image.

[0090] In this embodiment, the computer device inputs the gold standard for brain region segmentation of the sample brain image and the sample brain image into a second initial segmentation model, and obtains a sample mask map of the trigeminal nerve through the second initial segmentation model. Optionally, in this embodiment, the second initial segmentation model can be any one of the U-Net model, 3D U-Net model, V-Net model, and ResNet model.

[0091] Optionally, the computer device can obtain the above-mentioned sample brain images and the gold standard for brain region segmentation of the sample brain images from the PACS (Picture Archiving and Communication Systems) server. Alternatively, it can obtain the sample brain images in real time from the imaging device, and then perform segmentation processing on the sample brain images to obtain the gold standard for brain region segmentation of the sample brain images.

[0092] S502, based on the sample segmentation results and the gold standard for trigeminal nerve segmentation in the sample brain image, obtain the value of the first loss function of the second initial segmentation model.

[0093] Optionally, in this embodiment, the first loss function may include the Dice Loss function and the CL loss function. It should be noted that the combined use of the Dice Loss function and the CL loss function can not only emphasize the overall segmentation accuracy but also pay closer attention to the topological structure of the trigeminal nerve. Optionally, in this embodiment, the computer device can calculate the values ​​of the Dice Loss function and the CL loss function based on the sample segmentation results and the gold standard for trigeminal nerve segmentation in the sample brain image to obtain the value of the first loss function.

[0094] S503, the second initial segmentation model is trained based on the value of the first loss function to obtain the second segmentation model.

[0095] Optionally, in this embodiment, the computer device can adjust the parameters of the second initial segmentation model according to the value of the first loss function until the value of the first loss function reaches the minimum or stable value, and determine the second initial segmentation model corresponding to the value of the first loss function reaching the minimum or stable value as the aforementioned second segmentation model.

[0096] In this embodiment, by inputting the gold standard for brain region segmentation of the sample brain image and the sample brain image into the second initial segmentation model, the sample segmentation result of the trigeminal nerve of the sample brain image can be obtained. Thus, based on the sample segmentation result of the trigeminal nerve of the sample brain image and the gold standard for trigeminal nerve segmentation, the value of the first loss function of the second initial segmentation model can be accurately obtained. Then, based on the value of the first loss function, the second initial segmentation model can be accurately jointly trained to obtain a second segmentation model with high accuracy.

[0097] In the scenario described above, where the evaluation results corresponding to brain images are obtained through an evaluation network, the evaluation network needs to be pre-trained. In one embodiment, such as... Figure 9 As shown, the above method also includes:

[0098] S601, obtain the gold standard evaluation results corresponding to the sample brain images.

[0099] Optionally, in this embodiment, the computer device can obtain the gold standard evaluation result corresponding to the above-mentioned sample brain image from the PACS server, or it can analyze and evaluate the obtained sample brain image to obtain the gold standard evaluation result corresponding to the sample brain image. Optionally, the gold standard evaluation result may include the positive / negative classification result and the type of the positive category.

[0100] S602, input the sample brain image, the gold standard for segmentation of the trigeminal nerve in the sample brain image, and the sample clinical representation information corresponding to the sample brain image into the preset initial evaluation network to obtain the sample evaluation result corresponding to the sample brain image.

[0101] Optionally, in this embodiment, the initial evaluation network may include a first initial classification network and a second initial classification network. Optionally, the first initial classification network may be any one of the following networks: U-Net, 3D U-Net, V-Net, ResNet, etc.; the second initial classification network may be any one of the following networks: U-Net, 3D U-Net, V-Net, ResNet, etc. Optionally, in this embodiment, the sample brain image, the gold standard for trigeminal nerve segmentation of the sample brain image, and the corresponding clinical representation information of the sample brain image can be input into the first initial classification network to determine whether the sample brain image is a positive or negative category. Further, when the sample brain image is a positive category, the sample brain image, the gold standard for trigeminal nerve segmentation of the sample brain image, and the corresponding clinical representation information of the sample brain image can be input into the second initial classification network to determine the type of the sample brain image and obtain the sample evaluation result corresponding to the sample brain image.

[0102] S603. Based on the sample evaluation results and the gold standard evaluation results, the value of the second loss function of the initial evaluation network is obtained.

[0103] Optionally, the second loss function may include a Focal loss function and a CAM loss function. It should be noted that the joint loss function of Focal loss and CAM loss used in the initial evaluation network can further enhance the accuracy of the obtained sample evaluation results. Optionally, in this embodiment, the computer device can calculate the value of the joint loss function of Focal loss and CAM loss based on the values ​​of the sample evaluation results and the gold standard evaluation results, thereby obtaining the value of the second loss function of the initial evaluation network.

[0104] S604, The initial evaluation network is trained based on the value of the second loss function to obtain the evaluation network.

[0105] Optionally, in this embodiment, the computer device can adjust the parameters of the initial evaluation network according to the value of the second loss function, train the initial evaluation network until the value of the second loss function reaches the minimum or stable value, and obtain the above-mentioned evaluation network.

[0106] In this embodiment, by inputting the sample brain image, the gold standard for segmentation of the trigeminal nerve in the sample brain image, and the sample clinical representation information corresponding to the sample brain image into a preset initial evaluation network, the sample evaluation result corresponding to the sample brain image can be obtained. Thus, based on the sample evaluation result and the gold standard evaluation result, the value of the second loss function of the initial evaluation network can be obtained. Furthermore, the initial evaluation network can be accurately trained based on the value of the second loss function, and the evaluation network can be accurately obtained, thereby improving the accuracy of the obtained evaluation network.

[0107] To facilitate understanding by those skilled in the art, the brain image processing method provided in this application is described in detail below. This method may include:

[0108] S1, acquire sample brain images, the gold standard for brain region segmentation of sample brain images, and the gold standard for trigeminal nerve segmentation of sample brain images.

[0109] S2, input the sample brain image into the preset first initial segmentation model to obtain the sample segmentation result of the region of interest of the sample brain image.

[0110] S3. The first initial segmentation model is trained based on the sample segmentation results of the region of interest and the gold standard for brain region segmentation of the sample brain image to obtain the first segmentation model.

[0111] S4. Input the gold standard for brain region segmentation of the sample brain image and the sample brain image into the preset second initial segmentation model to obtain the sample segmentation result of the trigeminal nerve in the sample brain image.

[0112] S5. Based on the sample segmentation results and the gold standard for trigeminal nerve segmentation in the sample brain image, the value of the first loss function of the second initial segmentation model is obtained.

[0113] S6. Train the second initial segmentation model based on the value of the first loss function to obtain the second segmentation model.

[0114] S7, obtain the gold standard evaluation results corresponding to the sample brain images.

[0115] S8 inputs the sample brain image, the gold standard for segmentation of the trigeminal nerve in the sample brain image, and the sample clinical representation information corresponding to the sample brain image into the preset initial evaluation network to obtain the sample evaluation result corresponding to the sample brain image.

[0116] S9. Based on the sample evaluation results and the gold standard evaluation results, the value of the second loss function of the initial evaluation network is obtained.

[0117] S10, Train the initial evaluation network based on the value of the second loss function to obtain the evaluation network.

[0118] S11, determine the image type of the brain image through a preset discriminator; the image type is either a 2D image or a 3D image; if the brain image is a 2D image, input the brain image into the first segmentation sub-model to obtain the segmentation result of the region of interest of the brain image; or, if the brain image is a 3D image, input the brain image into the second segmentation sub-model to obtain the first mask map of the region of interest of the brain image; the region of interest is the brain region within a preset range around the trigeminal nerve in the brain image.

[0119] S12: Input the segmentation results of the region of interest and the brain image into the second segmentation model to obtain the segmentation results of the trigeminal nerve.

[0120] S13, input the brain image, the segmentation results of the trigeminal nerve, and the clinical representation information corresponding to the brain image into the first classification network in the evaluation network to determine the first classification result of the brain image; the first classification result is used to characterize whether the brain image is a positive or negative category.

[0121] S14, when the first classification result is a positive category, the brain image, segmentation result and corresponding clinical representation information of the brain image are input into the second classification network in the evaluation network to determine the second classification result of the brain image; the second classification result is used to characterize the type to which the brain image belongs, and the type to which the brain image belongs includes negative and the type of positive category to which the brain image belongs when it is a positive category.

[0122] It should be noted that the descriptions in S1-S14 above can be found in the relevant descriptions in the above embodiments, and their effects are similar, so they will not be repeated here.

[0123] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0124] Based on the same inventive concept, this application also provides a brain image processing apparatus for implementing the brain image processing method described above. The solution provided by this apparatus is similar to the implementation scheme described in the above method; therefore, the specific limitations in one or more brain image processing apparatus embodiments provided below can be found in the limitations of the brain image processing method described above, and will not be repeated here.

[0125] In one embodiment, such as Figure 10 As shown, a brain image processing device is provided, comprising: a first processing module and a second processing module, wherein:

[0126] The first processing module is used to input the brain image to be analyzed into a preset first segmentation model to obtain the segmentation result of the region of interest in the brain image; the region of interest is the brain region within a preset range around the trigeminal nerve in the brain image; the first segmentation model is obtained by training the preset first initial segmentation model according to the sample brain image and the gold standard for brain region segmentation of the sample brain image.

[0127] The second processing module is used to input the segmentation results of the region of interest and the brain image into a preset second segmentation model to obtain the segmentation results of the trigeminal nerve. The second segmentation model is obtained by training the preset second initial segmentation model based on the sample brain image, the gold standard for brain region segmentation of the sample brain image, and the gold standard for trigeminal nerve segmentation of the sample brain image.

[0128] The brain image processing device provided in this embodiment can execute the above method embodiment, and its implementation principle and technical effect are similar, so it will not be described again here.

[0129] Optionally, based on the above embodiments, the apparatus may further include:

[0130] The third processing module is used to input the brain image, segmentation results, and corresponding clinical representation information of the brain image into a preset evaluation network to obtain the evaluation results corresponding to the brain image; the evaluation results are used to characterize the feature information of the trigeminal nerve.

[0131] The brain image processing device provided in this embodiment can execute the above method embodiment, and its implementation principle and technical effect are similar, so it will not be described again here.

[0132] Based on the above embodiments, optionally, the above evaluation results include positive / negative classification results and the type of positive category; the above third processing module includes: a first determining unit and a second determining unit, wherein:

[0133] The first determining unit is used to input the brain image, segmentation result, and corresponding clinical representation information of the brain image into the first classification network in the evaluation network to determine the first classification result of the brain image; the first classification result is used to characterize the brain image as a positive or negative category.

[0134] The second determining unit is used to input the brain image, segmentation result, and corresponding clinical representation information of the brain image into the second classification network in the evaluation network when the first classification result is a positive category, and to determine the second classification result of the brain image; the second classification result is used to characterize the type to which the brain image belongs, and the type to which the brain image belongs includes negative and the type of positive category to which the brain image belongs when it is a positive category.

[0135] The brain image processing device provided in this embodiment can execute the above method embodiment, and its implementation principle and technical effect are similar, so it will not be described again here.

[0136] Based on the above embodiments, optionally, the segmentation result is a mask image of the trigeminal nerve; the above device further includes: a first acquisition module, wherein:

[0137] The first acquisition module is used to acquire a segmentation map of the trigeminal nerve based on the brain image and the mask map.

[0138] The brain image processing device provided in this embodiment can execute the above method embodiment, and its implementation principle and technical effect are similar, so it will not be described again here.

[0139] Based on the above embodiments, optionally, the first processing module includes: a third determining unit and an acquiring unit, wherein:

[0140] The third determining unit is used to determine the image type of the brain image through a preset discriminator; the image type is either a 2D image or a 3D image.

[0141] The acquisition unit is used to input the brain image into the first segmentation sub-model to obtain the first mask image if the brain image is a 2D image. Alternatively,

[0142] The acquisition unit is used to input the brain image into the second segmentation sub-model if the brain image is a 3D image, and obtain the first mask image.

[0143] The brain image processing device provided in this embodiment can execute the above method embodiment, and its implementation principle and technical effect are similar, so it will not be described again here.

[0144] Based on the above embodiments, optionally, the above apparatus further includes: a second acquisition module, a third acquisition module, and a first training module, wherein:

[0145] The second acquisition module is used to input the gold standard for brain region segmentation of the sample brain image and the sample brain image into the second initial segmentation model to obtain the sample segmentation result of the trigeminal nerve in the sample brain image.

[0146] The third acquisition module is used to obtain the value of the first loss function of the second initial segmentation model based on the sample segmentation results and the gold standard for segmentation of the trigeminal nerve in the sample brain image.

[0147] The first training module is used to train the second initial segmentation model based on the value of the first loss function to obtain the second segmentation model.

[0148] Optionally, the first loss function includes the Dice Loss function and the CL loss function.

[0149] The brain image processing device provided in this embodiment can execute the above method embodiment, and its implementation principle and technical effect are similar, so it will not be described again here.

[0150] Based on the above embodiments, optionally, the apparatus further includes: a fourth acquisition module, a fifth acquisition module, a sixth acquisition module, and a second training module, wherein:

[0151] The fourth acquisition module is used to acquire the gold standard evaluation results corresponding to the brain images of the samples.

[0152] The fifth acquisition module is used to input the sample brain image, the gold standard for segmentation of the trigeminal nerve in the sample brain image, and the sample clinical representation information corresponding to the sample brain image into a preset initial evaluation network to obtain the sample evaluation result corresponding to the sample brain image.

[0153] The sixth acquisition module is used to obtain the value of the second loss function of the initial evaluation network based on the sample evaluation results and the gold standard evaluation results.

[0154] The second training module is used to train the initial evaluation network based on the value of the second loss function to obtain the evaluation network.

[0155] Optionally, the second loss function includes the Focal loss function and the CAM loss function.

[0156] The brain image processing device provided in this embodiment can execute the above method embodiment, and its implementation principle and technical effect are similar, so it will not be described again here.

[0157] Each module in the aforementioned brain imaging processing device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the operations corresponding to each module.

[0158] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0159] The brain image to be analyzed is input into a preset first segmentation model to obtain the segmentation result of the region of interest in the brain image; the region of interest is the brain region within a preset range around the trigeminal nerve in the brain image; the first segmentation model is obtained by training the preset first initial segmentation model based on the sample brain image and the gold standard for brain region segmentation of the sample brain image.

[0160] The segmentation results of the region of interest and the brain image are input into the preset second segmentation model to obtain the segmentation results of the trigeminal nerve. The second segmentation model is obtained by training the preset second initial segmentation model based on the sample brain image and the gold standard for segmentation of the trigeminal nerve in the sample brain image.

[0161] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor:

[0162] The brain image to be analyzed is input into a preset first segmentation model to obtain the segmentation result of the region of interest in the brain image; the region of interest is the brain region within a preset range around the trigeminal nerve in the brain image; the first segmentation model is obtained by training the preset first initial segmentation model based on the sample brain image and the gold standard for brain region segmentation of the sample brain image;

[0163] The segmentation results of the region of interest and the brain image are input into the preset second segmentation model to obtain the segmentation results of the trigeminal nerve. The second segmentation model is obtained by training the preset second initial segmentation model based on the sample brain image and the gold standard for segmentation of the trigeminal nerve in the sample brain image.

[0164] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, performs the following steps:

[0165] The brain image to be analyzed is input into a preset first segmentation model to obtain the segmentation result of the region of interest in the brain image; the region of interest is the brain region within a preset range around the trigeminal nerve in the brain image; the first segmentation model is obtained by training the preset first initial segmentation model based on the sample brain image and the gold standard for brain region segmentation of the sample brain image;

[0166] The segmentation results of the region of interest and the brain image are input into the preset second segmentation model to obtain the segmentation results of the trigeminal nerve. The second segmentation model is obtained by training the preset second initial segmentation model based on the sample brain image and the gold standard for segmentation of the trigeminal nerve in the sample brain image.

[0167] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0168] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0169] The technical features of the above embodiments can be combined in any way. For the sake of brevity, 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.

[0170] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A brain image processing method, characterized in that, The method includes: The brain image to be analyzed is input into a preset first segmentation model to obtain the segmentation result of the region of interest in the brain image; the region of interest is the brain region within a preset range around the trigeminal nerve in the brain image; the first segmentation model is obtained by training a preset first initial segmentation model based on the sample brain image and the gold standard for brain region segmentation of the sample brain image. The segmentation results of the region of interest and the brain image are input into a preset second segmentation model to obtain the segmentation results of the trigeminal nerve. The second segmentation model is obtained by inputting the sample brain image and the gold standard for brain region segmentation of the sample brain image into a preset second initial segmentation model to obtain the sample segmentation results of the trigeminal nerve of the sample brain image. Based on the sample segmentation results and the gold standard for trigeminal nerve segmentation of the sample brain image, the value of the first loss function of the second initial segmentation model is obtained. The second initial segmentation model is trained based on the value of the first loss function. Obtain the gold standard evaluation result corresponding to the sample brain image, input the sample brain image, the gold standard segmentation of the trigeminal nerve in the sample brain image, and the sample clinical representation information corresponding to the sample brain image into a preset initial evaluation network to obtain the sample evaluation result corresponding to the sample brain image, obtain the value of the second loss function of the initial evaluation network based on the sample evaluation result and the gold standard evaluation result, and train the initial evaluation network based on the value of the second loss function to obtain the evaluation network; The brain image, the segmentation result of the trigeminal nerve, and the corresponding clinical representation information of the brain image are input into the first classification network of the evaluation network to determine a first classification result characterizing the brain image as a positive or negative category. When the first classification result is a positive category, the brain image, the segmentation result of the trigeminal nerve, and the corresponding clinical representation information of the brain image are input into the second classification network of the evaluation network to determine a second classification result of the brain image. The second classification result is used to characterize the type to which the brain image belongs, and the type to which the brain image belongs includes negative and the type of positive category to which the brain image belongs when it is a positive category.

2. The method according to claim 1, characterized in that, The segmentation result of the trigeminal nerve is a mask image of the trigeminal nerve; the method further includes: Based on the brain image and the mask image, a segmentation map of the trigeminal nerve is obtained.

3. The method according to claim 1 or 2, characterized in that, The step of inputting the brain image to be analyzed into a preset first segmentation model to obtain the segmentation result of the region of interest in the brain image includes: The image type of the brain image is determined by a preset discriminator; the image type is either a 2D image or a 3D image. If the brain image is a 2D image, then the brain image is input into the first segmentation sub-model to obtain the segmentation result of the region of interest; or, If the brain image is a 3D image, then the brain image is input into the second segmentation sub-model to obtain the segmentation result of the region of interest.

4. The method according to claim 3, characterized in that, The number of convolutional layers in the second segmentation sub-model is greater than the number of convolutional layers in the first segmentation sub-model.

5. The method according to claim 1, characterized in that, The first loss function includes the Dice Loss function and the CL loss function, and the second loss function includes the Focal loss function and the CAM loss function.

6. The method according to claim 1, characterized in that, The first segmentation model is any one of the U-Net model, V-Net model, and ResNet model.

7. The method according to claim 1, characterized in that, The second segmentation model includes, but is not limited to, the U-Net model, the 3D U-Net model, and the V-Net model.

8. The method according to claim 1, characterized in that, The evaluation network can be any one of the following: U-Net, V-Net, or ResNet.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 8.

Images