An artificial intelligence-based imaging analysis method and system for degenerative cervical spinal cord disease

Abstract

The application discloses an artificial intelligence-based image analysis method and system for degenerative cervical spinal cord disease, relates to the technical field of artificial intelligence, and solves the technical problems of the existing technology, such as non-uniform image report standards, rough image analysis, increased patient anxiety, wasted medical resources, significant observer bias in DCM evaluation, and low efficiency of degenerative cervical spinal cord disease image analysis. A cervical vertebra dissection segmentation model is used to obtain a segmentation label; an intervertebral canal stenosis quantification grading model is used to obtain an intervertebral canal stenosis prediction grade of the segmentation label for the intervertebral canal; a spinal cord compression classification model is used to obtain a spinal cord compression result of the segmentation label for the cervical spinal cord; and a cervical vertebra image analysis model is used to obtain an analysis report of the cervical vertebra nuclear magnetic image. The image report is standardized, improved segmentation models are used, classification models are used in cooperation, and cervical vertebra image analysis models are used for processing, so that the accuracy of analysis is improved, and the efficiency of degenerative cervical spinal cord disease image analysis is improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of artificial intelligence, and particularly relates to a degenerative cervical spinal cord disease image analysis method and system based on artificial intelligence. BACKGROUND

[0002] Degenerative cervical spinal cord disease (DCM) is a chronic and progressive cervical degenerative disease, and is the most common cause of chronic spinal cord injury in adults. This kind of disease mainly includes cervical spondylotic myelopathy, posterior longitudinal ligament ossification, yellow ligament ossification and intervertebral disc degeneration. With the growth of age, almost all people will have image manifestations of cervical degeneration, and early diagnosis is particularly important for preventing further damage to the spinal cord. Although clinical symptoms and signs are also important criteria for diagnosing DCM, the mechanism of human behavior adaptation and neural plasticity may mask the disease deterioration, so it is very meaningful to objectively evaluate the severity of spinal cord compression and spinal stenosis through image assessment.

[0003] At present, the diagnosis and treatment level of degenerative cervical spinal cord disease in community medical institutions is limited, and the differences in professional cognition of the disease of physicians in various units and the image report standards provided by image physicians lead to the lack of correct cognition of patients about DCM, and the situation of missed diagnosis or over-diagnosis often occurs. Rough image analysis often exacerbates the anxiety of patients, and also leads to the waste of medical resources and unnecessary burden on the cost of social medical insurance. At the same time, for large sample data research, the inter-observer bias of DCM is very large, which is not conducive to the development of multi-center clinical research of big data. Therefore, it is particularly important to develop a standardized and accurate deep learning algorithm to analyze cervical magnetic resonance images to assist clinical and image physicians in diagnosing cervical spinal stenosis caused by DCM and to provide standardized image analysis reports. SUMMARY

[0004] The present application aims to at least solve one of the technical problems existing in the prior art; for this purpose, the present application proposes a degenerative cervical spinal cord disease image analysis method and system based on artificial intelligence, which is used to solve the technical problems that the image report standards of the prior art are not unified, the rough image analysis increases the anxiety of patients and wastes medical resources, the DCM evaluation has significant inter-observer bias, and the efficiency of degenerative cervical spinal cord disease image analysis is low.

[0005] To achieve the above-mentioned purpose, the first aspect of the present application provides a degenerative cervical spinal cord disease image analysis system based on artificial intelligence, comprising a data acquisition module, a data analysis module and a database.

[0006] The data acquisition module: acquires cervical spine nuclear magnetic image through a data acquisition device;

[0007] The data analysis module: input the cervical vertebrae nuclear magnetic image into the cervical vertebrae anatomical segmentation model to obtain a segmentation label, the cervical vertebrae anatomical segmentation model is improved by a cross pseudo-supervised CPS framework on an nnU-Net architecture; input the detection image with the segmentation label of the spinal canal into the spinal canal stenosis quantification grading model to obtain a spinal canal stenosis prediction grade, the spinal canal stenosis quantification grading model is constructed by an artificial intelligence model; input the detection image with the segmentation label of the cervical spinal cord into the spinal cord compression classification model to obtain a spinal cord compression result, the spinal cord compression classification model is constructed by a ResNet network; input the detection image corresponding to the segmentation label, the spinal canal stenosis prediction grade and the spinal cord compression result into the cervical vertebrae image analysis model to obtain an analysis report, the cervical vertebrae image analysis model is constructed by a large language model.

[0008] The database is used for storing historical data required by the training model.

[0009] The application standardizes the image report through the above steps, processes the image report using the improved segmentation model, the classification model and the cervical vertebrae image analysis model, and gives an image analysis report, thereby providing an efficient and convenient image auxiliary analysis tool, and helping doctors to accurately diagnose degenerative cervical spinal cord disease and improving the analysis accuracy; the spinal canal stenosis and the spinal cord compression are graded and evaluated, which helps users to better evaluate the severity of intervertebral disc degeneration.

[0010] Further, the cervical vertebrae anatomical segmentation model is improved by a cross pseudo-supervised CPS framework on an nnU-Net architecture, comprising:

[0011] Obtain historical cervical vertebrae nuclear magnetic images and corresponding segmentation labels;

[0012] Divide the historical cervical vertebrae nuclear magnetic images and the corresponding segmentation labels into training data, verification data and test data; perform image preprocessing on the training data, the verification data and the test data to obtain a training set, a verification set and a test set;

[0013] Select an improved nnU-Net architecture as a base model;

[0014] Train the base model through the training set, and adjust the learning rate and the hyperparameters on the verification set to obtain a pre-trained model;

[0015] Verify the pre-trained model on the test set, and finally obtain a cervical vertebrae anatomical segmentation model with the cervical vertebrae nuclear magnetic image as the input and the segmentation label as the output.

[0016] Further, the base model is a neural network architecture based on U-Net and residual blocks, and is configured and trained through an nnU-Net framework, specifically comprising an encoder, a decoder, a skip connection module and a feature combination module.

[0017] Further, on the basis of the base model, the forced spacing setting, the optimizer and the preprocessing process are optimized, including:

[0018] Forced spacing setting: in the preprocessing stage, by modifying the spacing parameters of the original cervical spine magnetic resonance image, the patch size input into the model can be constrained at an optimal value;

[0019] Optimizer: Ranger optimizer is used to better adapt to the nnU-Net model improved by the semi-supervised learning idea through the CPS framework;

[0020] Preprocessing process: advanced data enhancement methods such as contrast adjustment and noise addition are added.

[0021] Further, the loss function used by the Ranger optimizer is obtained by the Dice loss and the cross-entropy loss, including:

[0022] Get the Dice loss L Dice And the cross-entropy loss L Cross-Entropy ;

[0023] The loss function L used by the Ranger optimizer is calculated by the formula;

[0024] L=L Dice +L Cross-Entropy +λ(L CPSlabeled +L CPSunlabeled ); Wherein, λ represents the weight, λ∈(0, 1); L CPSlabeled And L CPSunlabeled respectively represent the loss of labeled data and unlabeled data.

[0025] Further, the intervertebral canal stenosis quantification grading model is constructed by an artificial intelligence model, including:

[0026] Obtain a plurality of historical detection images whose segmentation labels are intervertebral canals and their corresponding intervertebral canal stenosis grades;

[0027] Divide the plurality of historical detection images and their corresponding intervertebral canal stenosis grades into training data, validation data and test data; image preprocessing is performed on the training data, validation data and test data to obtain a training set, a validation set and a test set;

[0028] Select an artificial intelligence model as a base model;

[0029] Train the base model through the training set, and adjust the learning rate and hyperparameters on the validation set to obtain a pre-trained model;

[0030] Through verification of the pre-training model on the test set, finally, the detection image with the segmentation label of the spinal canal is input, and the spinal canal stenosis quantification grading model with the spinal canal stenosis prediction grade as output is obtained.

[0031] Further, the spinal cord compression classification model is constructed through a ResNet network, comprising:

[0032] Obtain a plurality of historical detection images with the segmentation label of the cervical spinal cord and their corresponding spinal cord compression results;

[0033] Divide the plurality of historical detection images and their corresponding spinal cord compression results into training data, verification data and test data; and perform image preprocessing on the training data, verification data and test data to obtain a training set, a verification set and a test set;

[0034] Select a ResNet network model as a base model;

[0035] Train the base model through the training set, and adjust the learning rate and hyperparameters on the verification set to obtain a pre-training model;

[0036] Through verification of the pre-training model on the test set, finally, the detection image with the segmentation label of the cervical spinal cord is input, and the spinal cord compression classification model with the spinal cord compression result as output is obtained.

[0037] Further, the ResNet network model adopts a ResNet network model with at least N layers to perform morphological analysis on the detection image with the segmentation label of the cervical spinal cord through multi-layer feature extraction and multi-scale feature fusion to obtain the spinal cord compression result of compression or not, compression position and compression degree; wherein N is a positive integer.

[0038] Further, the cervical vertebra image analysis model is constructed through a large language model, comprising:

[0039] Obtain a plurality of historical detection images, historical spinal canal stenosis prediction grades and historical spinal cord compression results corresponding to the segmentation label and their corresponding historical analysis reports;

[0040] Divide the plurality of historical detection images, historical spinal canal stenosis prediction grades and historical spinal cord compression results corresponding to the segmentation label and their corresponding historical analysis reports into training data, verification data and test data; and perform data preprocessing on the training data, verification data and test data to obtain a training set, a verification set and a test set;

[0041] Select a large language network model as a base model;

[0042] Train the base model through the training set, and adjust the learning rate and hyperparameters on the verification set to obtain a pre-training model;

[0043] By verifying the pre-training model on the test set, finally, the cervical image analysis model with the input of a plurality of detection images corresponding to the segmentation label, the spinal canal stenosis prediction level and the spinal cord compression result, and the output of the analysis report is obtained.

[0044] Another aspect of the present application provides an artificial intelligence-based degenerative cervical spinal cord disease image analysis method, comprising:

[0045] S0: acquiring a cervical nucleus magnetic image;

[0046] S1: inputting the cervical nucleus magnetic image into a cervical dissection segmentation model to obtain a segmentation label;

[0047] S2: inputting the detection image of the segmentation label for the intervertebral canal into an intervertebral canal stenosis quantification grading model to obtain a spinal canal stenosis prediction level;

[0048] S3: inputting the detection image of the segmentation label for the cervical spinal cord into a spinal cord compression classification model to obtain a spinal cord compression result;

[0049] S4: inputting a plurality of detection images corresponding to the segmentation label, the spinal canal stenosis prediction level and the spinal cord compression result into a cervical image analysis model to obtain an analysis report.

[0050] Compared with the prior art, the present application has the following beneficial effects:

[0051] 1. The present application inputs the cervical nucleus magnetic image into the cervical dissection segmentation model to obtain the segmentation label; inputs the detection image of the segmentation label for the intervertebral canal into the intervertebral canal stenosis quantification grading model to obtain the spinal canal stenosis prediction level; inputs the detection image of the segmentation label for the cervical spinal cord into the spinal cord compression classification model to obtain the spinal cord compression result; and inputs a plurality of detection images corresponding to the segmentation label, the spinal canal stenosis prediction level and the spinal cord compression result into the cervical image analysis model to obtain the analysis report. The image report is standardized, an improved segmentation model is used, a classification model and a cervical image analysis model are used for processing, the analysis accuracy is improved, and the efficiency of degenerative cervical spinal cord disease image analysis is improved.

[0052] 2. The present application uses an improved nnU-Net network based on a semi-supervised deep learning framework of CPS idea, and adapts a specific optimizer. The combination of the model and the optimizer improves the segmentation and recognition accuracy of the model, and the semi-supervised training method improves the performance of the model under limited labeling.

[0053] 3、The application generates an analysis report by using a pre-trained large language model, can utilize the basic knowledge and question answering capabilities of the large language model, expand services, provide personalized analysis and guidance for nuclear magnetic imaging for users, and at the same time realize standardized analysis results, provide support for large sample data research and multi-center clinical research. BRIEF DESCRIPTION OF DRAWINGS

[0054] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiment or prior art description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and those skilled in the art can also obtain other drawings according to these drawings without creative labor.

[0055] Figure 1 A principle schematic diagram of an artificial intelligence-based degenerative cervical spinal cord disease image analysis system of the present application;

[0056] Figure 2 A flowchart of an artificial intelligence-based degenerative cervical spinal cord disease image analysis method of the present application. DETAILED DESCRIPTION

[0057] The technical solutions of the present application will be described below in conjunction with the embodiments, obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0058] Please refer to Figure 1 The first aspect embodiment of the present application provides an artificial intelligence-based degenerative cervical spinal cord disease image analysis system, which comprises a data acquisition module, a data analysis module and a database.

[0059] Data acquisition module: acquire cervical spine nuclear magnetic imaging through data acquisition equipment; the data acquisition equipment includes various sensors and the like;

[0060] The data analysis module: input the cervical spine MRI image into the cervical spine anatomical segmentation model to obtain the segmentation label, the cervical spine anatomical segmentation model is improved through the cross pseudo-supervised CPS framework on the nnU-Net architecture, and the segmentation label refers to the name after the cervical spine MRI image is segmented, including cervical spine body, intervertebral disc, cervical spinal cord, cerebrospinal fluid and paravertebral muscle tissue, etc.; the detection image with the segmentation label of the spinal canal is input into the spinal canal stenosis quantification grading model to obtain the spinal canal stenosis prediction grade, the spinal canal stenosis quantification grading model is constructed through an artificial intelligence model, and the spinal canal stenosis prediction grade refers to the degree of spinal canal stenosis, which can be divided into four grades of 1, 2, 3 and 4 in this embodiment; the detection image with the segmentation label of the cervical spinal cord is input into the spinal cord compression classification model to obtain the spinal cord compression result, the spinal cord compression classification model is constructed through a ResNet network, and the spinal cord compression result includes two kinds of compression and no compression, and the compression includes three kinds of front compression, rear compression and front and rear compression; the analysis report is obtained by inputting the segmentation label corresponding to the plurality of detection images, the spinal canal stenosis prediction grade and the spinal cord compression result into the cervical spine image analysis model, the cervical spine image analysis model is constructed through a large language model, and the analysis report is a comprehensive analysis of the cervical spine MRI image.

[0061] The database is used to store a plurality of historical data required for training the model.

[0062] The cervical spine anatomical segmentation model in this embodiment is improved through the cross pseudo-supervised CPS framework on the nnU-Net architecture, including:

[0063] A plurality of historical cervical spine MRI images and their corresponding segmentation labels are obtained.

[0064] The plurality of historical cervical spine MRI images and their corresponding segmentation labels are divided into training data, verification data and test data; image preprocessing is performed on the training data, verification data and test data to obtain a training set, a verification set and a test set; the ratio between the training set, the test set and the verification set is 7:2:1;

[0065] The improved nnU-Net architecture is selected as the base model.

[0066] The base model is trained through the training set, and the learning rate and hyperparameters are adjusted on the verification set to obtain a pre-trained model.

[0067] The pre-trained model is verified on the test set, and finally the cervical spine anatomical segmentation model with the cervical spine MRI image as the input and the segmentation label as the output is obtained.

[0068] The base model in this embodiment is a neural network architecture based on U-Net and residual blocks, and is configured and trained through the nnU-Net framework, specifically including an encoder, a decoder, a jump connection module and a feature combination module.

[0069] In this embodiment, the forced spacing setting, optimizer and preprocessing process are optimized on the basis of the basic model, including:

[0070] Forced spacing setting: in the preprocessing stage, by modifying the spacing parameters of the original cervical spine magnetic resonance image, the patch size input into the model can be constrained to an optimal value; by adjusting the spacing parameters, the input patch size is neither too large nor too small, reaching a balance point, ensuring that the input patch size is constrained to the optimal value, reducing the computational burden of the model in the inference stage, not only optimizing the inference speed and improving the segmentation efficiency, but also reducing the resource consumption;

[0071] Optimizer: Ranger optimizer is used to better adapt to the nnU-Net model improved by the semi-supervised learning idea through the CPS framework;

[0072] Preprocessing process: advanced data enhancement methods such as contrast adjustment and noise addition are added, and the advanced data enhancement methods such as contrast adjustment and noise addition are added in this embodiment to further improve the adaptability of the model to different types of images, and for input images of different resolutions, a multi-scale processing method is used to improve the detection ability of the model for different size lesion regions.

[0073] The loss function used by the Ranger optimizer in this embodiment is obtained from the Dice loss and the cross-entropy loss, including:

[0074] Get the Dice loss L Dice and the cross-entropy loss L Cross-Entropy ; wherein the Dice loss and the cross-entropy loss are configured by default by nnU-Net, which has been proven to be robust in medical image segmentation tasks;

[0075] The loss function L used by the Ranger optimizer is calculated by the formula;

[0076] L=L Dice +L Cross-Entropy +λ(L CPSlabeled +L CPSunlabeled ); wherein λ represents the weight, λ∈(0, 1), and the specific value is set according to experience; L CPSlabeled and L CPSunlabeled represent the loss of labeled data and unlabeled data respectively.

[0077] Through the above steps, the semi-supervised deep learning architecture of the CPS concept is used to optimize the nnU-Net network, and the adaptive optimizer is configured, through the synergistic effect of the model and the optimizer, the segmentation and recognition accuracy of the model is enhanced, and with the help of the semi-supervised training strategy, the performance of the model is effectively improved under the condition of limited labeled data.

[0078] The intervertebral canal stenosis quantification grading model in the embodiment is constructed by an artificial intelligence model, and includes:

[0079] A plurality of historical detection images with an intervertebral canal segmentation label and corresponding intervertebral canal stenosis grades are obtained.

[0080] The plurality of historical detection images and the corresponding intervertebral canal stenosis grades are divided into training data, verification data and test data; the training data, the verification data and the test data are preprocessed to obtain a training set, a verification set and a test set; the ratio between the training set, the test set and the verification set is 7:2:1;

[0081] An artificial intelligence model is selected as a base model; the artificial intelligence model includes a convolutional neural network model and the like.

[0082] The base model is trained by the training set, and a learning rate and hyperparameters are adjusted on the verification set to obtain a pre-trained model.

[0083] The pre-trained model is verified on the test set, and finally an intervertebral canal stenosis quantification grading model with an input of a detection image with an intervertebral canal segmentation label and an output of an intervertebral canal stenosis prediction grade is obtained.

[0084] The intervertebral canal stenosis grade in the embodiment is obtained by a plurality of discipline experts marking cervical spine magnetic resonance images; the marking standard of the embodiment is that at least 2 cervical spine chief physicians and at least 1 imaging physician make a judgment on the cervical spine canal stenosis grade of the corresponding segment C2 / 3-C7 / T1 under the same type of machine in the reading window, and the judgment result consistent with more than 2 / 3 experts is selected as the gold standard; the image reading process is operated based on the window, and there is no mutual interference between the readers; in the process of marking the cervical spine sagittal position, there is a corresponding transverse section to assist the reader to interpret the cervical spine magnetic resonance image; the image data has been desensitized before the cervical spine magnetic resonance image is interpreted.

[0085] The spinal cord compression classification model in the embodiment is constructed by a ResNet network, and includes:

[0086] A plurality of historical detection images with a cervical spinal cord segmentation label and corresponding spinal cord compression results are obtained; the spinal cord compression result is a compression result determined by a related expert based on the cerebrospinal fluid area and the curvature;

[0087] The plurality of historical detection images and the corresponding spinal cord compression results are divided into training data, verification data and test data; the training data, the verification data and the test data are preprocessed to obtain a training set, a verification set and a test set; the ratio between the training set, the test set and the verification set is 7:2:1;

[0088] The ResNet network model is selected as the base model;

[0089] The base model is trained through the training set, and the pre-training model is obtained by adjusting the learning rate and hyperparameters on the validation set;

[0090] The pre-training model is verified on the test set, and finally the cervical spine image analysis model is obtained, which inputs the segmentation label corresponding to several detection images, the predicted grade of spinal canal stenosis, and the spinal cord compression result, and outputs the analysis report.

[0091] Through the above steps, not only the cervical nucleus magnetic image is recognized and the cervical degenerative change is recognized, but also the judgment of the spinal canal stenosis grade and the judgment of the compression situation and the compression position are realized, which realizes the more comprehensive analysis effect of the image by using artificial intelligence, and improves the efficiency of the degenerative cervical spinal cord disease image analysis.

[0092] The ResNet network model in the present embodiment adopts a ResNet network model with at least N layers to perform morphological analysis on the detection image with the segmentation label of cervical spinal cord to obtain the spinal cord compression result of whether there is compression, the compression position and the compression degree; wherein N is a positive integer, and in the present embodiment, N is set to 30.

[0093] The cervical image analysis model in the present embodiment is constructed by a large language model, which includes:

[0094] Obtain several historical detection images corresponding to the segmentation label, historical spinal canal stenosis prediction grades, and historical spinal cord compression results and their corresponding historical analysis reports;

[0095] Divide the several historical detection images corresponding to the segmentation label, the historical spinal canal stenosis prediction grades, and the historical spinal cord compression results and their corresponding historical analysis reports into training data, validation data and test data; perform data preprocessing on the training data, validation data and test data to obtain a training set, a validation set and a test set; the ratio between the training set, the test set and the validation set is 7:2:1;

[0096] Select a large language network model as the base model; the large language model includes Llama-3-8B model, etc.

[0097] The base model is trained through the training set, and the pre-training model is obtained by adjusting the learning rate and hyperparameters on the validation set;

[0098] The pre-training model is verified on the test set, and finally the cervical spine image analysis model is obtained, which inputs the segmentation label corresponding to several detection images, the predicted grade of spinal canal stenosis, and the spinal cord compression result, and outputs the analysis report.

[0099] Please refer toFigure 2 Another aspect of the present application provides an artificial intelligence-based image analysis method for degenerative cervical myelopathy, comprising:

[0100] S0: acquiring a cervical spine MRI image;

[0101] S1: inputting the cervical spine MRI image into a cervical spine anatomical segmentation model to obtain a segmentation label;

[0102] S2: inputting a detection image with the segmentation label as an intervertebral canal into an intervertebral canal stenosis quantification grading model to obtain an intervertebral canal stenosis prediction grade;

[0103] S3: inputting a detection image with the segmentation label as a cervical spinal cord into a spinal cord compression classification model to obtain a spinal cord compression result;

[0104] S4: inputting a plurality of detection images corresponding to the segmentation label, the intervertebral canal stenosis prediction grade, and the spinal cord compression result into a cervical spine image analysis model to obtain an analysis report.

[0105] Some data in the above formula are calculated by removing the dimension, and the formula is obtained by software simulation of a large amount of collected data to obtain a formula closest to the actual situation; the preset parameters and the preset threshold in the formula are set by a person skilled in the art according to the actual situation or obtained by a large amount of data simulation.

[0106] The working principle of the present application is as follows: a cervical spine MRI image is acquired; the cervical spine MRI image is input into a cervical spine anatomical segmentation model to obtain a segmentation label; a detection image with the segmentation label as an intervertebral canal is input into an intervertebral canal stenosis quantification grading model to obtain an intervertebral canal stenosis prediction grade; a detection image with the segmentation label as a cervical spinal cord is input into a spinal cord compression classification model to obtain a spinal cord compression result; a plurality of detection images corresponding to the segmentation label, the intervertebral canal stenosis prediction grade, and the spinal cord compression result are input into a cervical spine image analysis model to obtain an analysis report. The image report is standardized, and an improved segmentation model, a classification model, and a cervical spine image analysis model are used for processing, which improves the accuracy of analysis and further improves the efficiency of image analysis for degenerative cervical myelopathy, avoids the problems of non-uniform image report standards in the prior art, rough image analysis, increased patient anxiety, and waste of medical resources, and the significant observer bias in DCM judgment, resulting in low efficiency of image analysis for degenerative cervical myelopathy.

[0107] The above embodiments are only used to illustrate the technical method of the present application and not to limit it. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present application can be modified or replaced equivalently without departing from the spirit and scope of the technical method of the present application.​

Claims

1. An artificial intelligence-based image analysis system for degenerative cervical spinal cord disease, characterized by, The application relates to a cervical vertebra image analysis method and device. The method comprises the following steps: The data acquisition module acquires cervical vertebra MRI images through a data acquisition device; The data analysis module inputs the cervical vertebra MRI images into a cervical vertebra anatomical segmentation model to obtain segmentation labels; inputs the segmentation labels into an intervertebral canal stenosis quantification grading model to obtain a spinal canal stenosis prediction grade; inputs the segmentation labels into a spinal cord compression classification model to obtain a spinal cord compression result; and inputs the segmentation labels, the spinal canal stenosis prediction grade and the spinal cord compression result into a cervical vertebra image analysis model to obtain an analysis report; wherein the cervical vertebra anatomical segmentation model is obtained by improving an nnU-Net architecture through a cross pseudo-supervision CPS framework; the intervertebral canal stenosis quantification grading model is constructed through an artificial intelligence model; the spinal cord compression classification model is constructed through a ResNet network; and the cervical vertebra image analysis model is constructed through a large language model. The cervical vertebra anatomical segmentation model is obtained by improving an nnU-Net architecture through a cross pseudo-supervision CPS framework, and comprises the following steps: A plurality of historical cervical vertebra MRI images and corresponding segmentation labels are acquired; The plurality of historical cervical vertebra MRI images and corresponding segmentation labels are divided into training data, verification data and test data; and the training data, the verification data and the test data are subjected to image preprocessing to obtain a training set, a verification set and a test set; An improved nnU-Net architecture is selected as a basic model; The basic model is trained through the training set, and a pre-training model is obtained by adjusting a learning rate and hyperparameters on the verification set; The pre-training model is verified on the test set, and finally a cervical vertebra anatomical segmentation model with the cervical vertebra MRI images as input and the segmentation labels as output is obtained; The basic model is a neural network architecture based on a U-Net and a residual block, and is configured and trained through an nnU-Net framework, and specifically comprises an encoder, a decoder, a jump connection module and a feature combination module; On the basis of the basic model, a forced spacing setting, an optimizer and a preprocessing process are optimized, comprising: The forced spacing setting: in the preprocessing stage, the spacing parameters of the original cervical vertebra MRI images are modified, so that the patch size input into the model can be constrained at an optimal value; The optimizer: a Ranger optimizer is used to better adapt to the nnU-Net model improved through the CPS framework semi-supervised learning idea; The preprocessing process: advanced data enhancement methods such as contrast adjustment and noise addition are added.

2. The artificial intelligence-based imaging analysis system for degenerative cervical myelopathy according to claim 1, characterized by, The loss function used by the Ranger optimizer is obtained from a Dice loss and a cross-entropy loss, comprising: Get dice loss and cross entropy loss ; The loss function L used by the Ranger optimizer is calculated through a formula; ; wherein, represents a weight, ∈(0, 1); and respectively represent the loss of labeled data and unlabeled data.

3. The artificial intelligence-based imaging analysis system for degenerative cervical myelopathy according to claim 1, characterized by, The intervertebral canal stenosis quantification grading model is constructed through an artificial intelligence model, comprising: A plurality of historical detection images with the segmentation labels being intervertebral canals and corresponding intervertebral canal stenosis grades are acquired; The historical detection images and corresponding intervertebral canal stenosis grades are divided into training data, verification data and test data; the training data, verification data and test data are preprocessed to obtain a training set, a verification set and a test set; An artificial intelligence model is selected as a base model; The base model is trained by using the training set, and a pre-training model is obtained by adjusting a learning rate and hyperparameters on the verification set; The pre-training model is verified on the test set, and finally an intervertebral canal stenosis quantification grading model is obtained, wherein the input is a detection image with a segmentation label of an intervertebral canal, and the output is an intervertebral canal stenosis prediction grade.

4. The artificial intelligence-based imaging analysis system for degenerative cervical myelopathy according to claim 1, characterized by, The spinal cord compression classification model is constructed by using a ResNet network, and includes: A plurality of historical detection images with a segmentation label of a cervical spinal cord and corresponding spinal cord compression results are obtained; The historical detection images and corresponding spinal cord compression results are divided into training data, verification data and test data; the training data, verification data and test data are preprocessed to obtain a training set, a verification set and a test set; A ResNet network model is selected as a base model; The base model is trained by using the training set, and a pre-training model is obtained by adjusting a learning rate and hyperparameters on the verification set; The pre-training model is verified on the test set, and finally a spinal cord compression classification model is obtained, wherein the input is a detection image with a segmentation label of a cervical spinal cord, and the output is a spinal cord compression result.

5. The artificial intelligence-based imaging analysis system for degenerative cervical myelopathy according to claim 4, wherein The ResNet network model adopts a ResNet network model with at least N layers to perform morphological analysis on the detection image with the segmentation label of the cervical spinal cord by multi-layer feature extraction and multi-scale feature fusion to obtain the spinal cord compression result of whether compression, compression position and compression degree; wherein N is a positive integer.

6. The artificial intelligence-based imaging analysis system for degenerative cervical myelopathy according to claim 1, wherein The cervical vertebra image analysis model is constructed by using a large language model, and includes: A plurality of historical detection images, historical intervertebral canal stenosis prediction grades and historical spinal cord compression results corresponding to a segmentation label and corresponding historical analysis reports are obtained; The historical detection images, historical intervertebral canal stenosis prediction grades and historical spinal cord compression results corresponding to the segmentation label and corresponding historical analysis reports are divided into training data, verification data and test data; the training data, verification data and test data are preprocessed to obtain a training set, a verification set and a test set; A large language network model is selected as a base model; The base model is trained by using the training set, and a pre-training model is obtained by adjusting a learning rate and hyperparameters on the verification set; The pre-training model is verified on the test set, and finally a cervical vertebra image analysis model is obtained, wherein the input is a plurality of detection images corresponding to the segmentation label, intervertebral canal stenosis prediction grades and spinal cord compression results, and the output is an analysis report.

7. An artificial intelligence-based image analysis method for degenerative cervical spinal cord disease, applied to the artificial intelligence-based image analysis system for degenerative cervical spinal cord disease according to any one of claims 1-6, characterized in that, It includes: S0: obtaining a cervical vertebra MRI image; S1: inputting the cervical vertebra MRI image into a cervical vertebra dissection segmentation model to obtain a segmentation label; S2: inputting the detection image with the segmentation label of the intervertebral canal into an intervertebral canal stenosis quantification grading model to obtain an intervertebral canal stenosis prediction grade; S3: inputting the detection image with the segmentation label of the cervical spinal cord into a spinal cord compression classification model to obtain a spinal cord compression result; S4: inputting the segmentation label corresponding to the several detection images, the spinal canal stenosis prediction grade and the spinal cord compression result into the cervical vertebra image analysis model to obtain an analysis report.

Images