Deep learning driven three-dimensional medical image diagnosis auxiliary system

Abstract

The invention discloses a deep learning-driven three-dimensional medical image diagnosis auxiliary system, in which a model training module is used as a core of the deep learning-driven three-dimensional medical image diagnosis auxiliary system, and the elaborate design and cooperative work of internal sub-modules of the model training module bring remarkable beneficial effects to the whole system. The data loading sub-module ensures efficient reading and preprocessing of large-scale image data, and provides a stable data stream for subsequent training. According to the model training module, through automatic feature learning and model customization and optimization, the system realizes efficient conversion from original image data to diagnosis knowledge, and the accuracy and efficiency of diagnosis are greatly improved. A doctor can make a treatment scheme more quickly according to an accurate diagnosis result provided by the system, and the prognosis of a patient is improved. And meanwhile, the stability and reliability of the model in the whole training process are ensured through efficient training management and performance monitoring and evaluation.

Description

Technical Field

[0001] The present invention belongs to the field of three-dimensional medical imaging technology, and specifically is a three-dimensional medical imaging diagnosis auxiliary system driven by deep learning. Background Art

[0002] Three-dimensional medical imaging diagnosis mainly uses computed tomography (CT), magnetic resonance imaging (MRI), ultrasound imaging and other equipment to obtain three-dimensional images of tissues in the patient's body. With the continuous advancement of medical technology, three-dimensional medical images such as MRI, CT and PET are playing an increasingly important role in clinical diagnosis. However, traditional medical imaging diagnosis mainly relies on the doctor's experience and visual judgment, and has problems such as strong subjectivity, low efficiency, and easy missed diagnosis. In order to improve the accuracy and efficiency of diagnosis, deep learning technology has been widely used in the field of medical image analysis in recent years. As an advanced medical diagnostic technology, three-dimensional medical imaging diagnosis is widely used in clinical diagnosis, disease treatment and medical research.

[0003] However, the feature extraction in the existing technology is inaccurate and relies on manually designed features, which makes it difficult to fully capture the deep information in complex three-dimensional images; the model training is difficult and prone to overfitting or underfitting. At the same time, the diagnosis prediction in the existing technology is not flexible and difficult to adapt to the diagnosis needs of various diseases; the result display is not intuitive and lacks intuitiveness and interactivity. Summary of the invention

[0004] The purpose of the present invention is to provide a deep learning driven three-dimensional medical imaging diagnosis auxiliary system to solve the above-mentioned problems.

[0005] The technical solution adopted by the present invention is as follows: a deep learning-driven three-dimensional medical imaging diagnosis auxiliary system, the system comprising: an image data acquisition module, a data preprocessing module, a feature extraction module, a model training module, a diagnosis prediction module, and a result display module;

[0006] The model training module is internally provided with a data preparation submodule, a model design submodule, a training process management submodule, a performance evaluation submodule, a hyperparameter adjustment submodule and a model saving and loading submodule;

[0007] The image data acquisition module is responsible for acquiring the original three-dimensional image data and corresponding annotation information from the medical imaging device, and these data are then transmitted to the data preprocessing module; the data preprocessing module performs denoising, alignment, correction, normalization, cropping or scaling operations on the original image data, and converts it into a format that meets the model input requirements and stores it for subsequent use; the preprocessed data enters the feature extraction module;

[0008] The feature extraction module automatically extracts key features in the image through three-dimensional convolutional neural network deep learning technology, and these features will be used to describe the structure and pattern in the image;

[0009] The model training module features and corresponding annotated data are used to train the model, and the model parameters are adjusted through the optimization algorithm to minimize the prediction error; the trained model is used in the diagnosis prediction module to predict new image data and generate diagnosis results;

[0010] The result display module presents the prediction results of the model to doctors and patients in an intuitive manner, including lesion annotations on three-dimensional images, prediction probabilities, and detailed diagnostic reports. It also displays changes in performance indicators during model training, thereby completing the entire process from data collection to result display.

[0011] In a preferred embodiment, the image data acquisition module is responsible for acquiring original three-dimensional image data from various medical imaging devices; using a magnetic resonance imaging device to scan the patient's brain to generate a series of two-dimensional slices, which are then synthesized into a three-dimensional image; during the data acquisition process, it is necessary to ensure that the resolution, contrast and signal-to-noise ratio of the image meet the requirements of subsequent processing; in addition, it is also necessary to collect patient information corresponding to the image, including age, gender, and clinical diagnosis, which will be used as annotation data together with the image data for subsequent analysis and model training; after the data acquisition is completed, the image data and annotation data are stored in a secure database for subsequent access and processing.

[0012] In a preferred embodiment, the data preprocessing module first performs denoising on the three-dimensional image to reduce the noise introduced by equipment or environmental factors; then, the image is aligned and corrected to ensure the spatial consistency of images of different patients; in addition, the image is normalized to scale the pixel values ​​to a uniform range, such as 0 to 1, to eliminate the differences in brightness and contrast between different images; the preprocessing also includes cropping or scaling the image to match the size requirements of the model input; finally, data enhancement operations are performed as needed, including random rotation, flipping or scaling, to increase the generalization ability of the model.

[0013] In a preferred embodiment, the data preparation submodule specifically includes:

[0014] Data loading: Load the preprocessed 3D image data X and the corresponding annotation data Y;

[0015] X: A tensor of shape (N,C,D,H,W), where N is the number of samples, C is the number of channels, and D, H, and W are depth, height, and width, respectively.

[0016] Y: A tensor of shape (N, L), where L is the number of labels;

[0017] Data enhancement: Randomly rotate, scale, and flip X to generate enhanced data X_aug;

[0018] Data splitting: Split X_aug and Y into training set (X_train, Y_train), validation set (X_val, Y_val) and test set (X_test, Y_test).

[0019] In a preferred embodiment, the model design submodule specifically includes:

[0020] Network architecture selection: Select 3DCNN as the basic architecture;

[0021] Layer and parameter configuration:

[0022] Input layer: accepts input of shape (C, D, H, W);

[0023] Convolutional layer: Use multiple 3D convolutional layers, and the parameters of each convolutional layer include convolution kernel size k, stride s, and padding p;

[0024] Activation function: Use ReLU function;

[0025] Pooling layer: Use 3D maximum pooling layer, parameters include pooling kernel size k_pool and step size s_pool;

[0026] Fully connected layer: Flatten the feature map and connect it to the fully connected layer, and finally output the probabilities of L categories;

[0027] Optimizer selection: Select the Adam optimizer, and the parameters include learning rate α and weight decay λ.

[0028] In a preferred embodiment, the training process management submodule specifically includes:

[0029] Batch processing: X_train and Y_train are processed in batches, each batch size is B;

[0030] Forward propagation: For each batch of data X_batch, the output Y_pred is calculated by 3DCNN;

[0031] Y_pred = CNN(X_batch)

[0032] Loss calculation

[0033] Use the cross entropy loss function to calculate the loss L;

[0034] L=-Σ(Y_true*log(Y_pred)) / B

[0035] Where Y_true is the true label and Y_pred is the probability distribution predicted by the model;

[0036] Back Propagation

[0037] Calculate the gradient of the loss function with respect to the network parameters;

[0038] Parameter Update

[0039] Use Adam optimizer to update network parameters;

[0040]

[0041] Where θ is the network parameter, α is the learning rate, is the gradient of the loss function with respect to θ.

[0042] In a preferred embodiment, the performance evaluation submodule specifically includes:

[0043] Indicator calculation: calculate accuracy Acc, recall Rec, and F1 score F1;

[0044] Acc=(TP+TN) / (P+N)

[0045] Rec=TP / P

[0046] F1=2*(Prec*Rec) / (Prec+Rec)

[0047] Where TP is the true positive example, TN is the true negative example, P is the total number of positive examples, N is the total number of negative examples, and Prec is the precision;

[0048] Validation set test: Test the model performance on the validation set and calculate the above indicators;

[0049] Model selection: Select the best model based on the performance indicators on the validation set;

[0050] The hyperparameter adjustment submodule specifically includes:

[0051] Hyperparameter search: Use grid search or random search methods to find the best hyperparameter combination, including learning rate α and batch size B;

[0052] Cross-validation: Use K-fold cross-validation to evaluate model performance under different hyperparameter combinations;

[0053] The model saving and loading submodule saves the trained model parameters to a file for subsequent use.

[0054] In a preferred embodiment, in the feature extraction module, the collected raw image data is preprocessed and feature extracted; first, the three-dimensional image is preprocessed by denoising, alignment and normalization to reduce the impact of noise and variability; then, a specific algorithm is used to extract useful features from the preprocessed image; these features include morphological features, texture features and intensity features; the purpose of feature extraction is to convert high-dimensional image data into a lower-dimensional feature representation that is easier for the model to process; the extracted features will be stored and used for subsequent model training.

[0055] In a preferred embodiment, in the diagnosis prediction module, the trained model is used to perform diagnosis prediction on new image data; when new patient image data is input, the image is first subjected to the same preprocessing and feature extraction operations, and then the extracted features are input into the trained 3DCNN model; the model calculates the output through forward propagation to obtain the probability that the patient belongs to different diagnostic categories; based on these probabilities, the optimal diagnostic result for the patient is determined; if the model outputs that the probability that the patient has a certain disease exceeds a set threshold, the disease is diagnosed; the diagnosis prediction results will be recorded and used for subsequent clinical decision-making.

[0056] In a preferred embodiment, in the result display module, the diagnostic prediction results of the model are displayed to doctors and patients in an intuitive and easy-to-understand manner; a graphical user interface is used to display the patient's three-dimensional image, and the lesion area predicted by the model is marked on the image; at the same time, a detailed diagnostic report is provided, including the model's predicted probability, disease type, and recommended further examinations and treatment plans; in addition, visualization tools are used to display the loss changes and performance indicator changes during the model training process, so that doctors and researchers can evaluate the performance and stability of the model; the goal of the result display module is to provide comprehensive and accurate information to assist doctors in making better clinical decisions.

[0057] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0058] 1. In the present invention, the model training module is the core of the deep learning-driven three-dimensional medical imaging diagnosis auxiliary system, and the careful design and collaborative work of its internal submodules bring significant beneficial effects to the entire system. The data loading submodule ensures the efficient reading and preprocessing of large-scale image data, and provides a stable data stream for subsequent training. Through automated feature learning and model customization and optimization, the model training module realizes the efficient conversion from raw image data to diagnostic knowledge, greatly improving the accuracy and efficiency of diagnosis. Doctors can rely on the accurate diagnosis results provided by the system to formulate treatment plans more quickly and improve patient prognosis. At the same time, efficient training management and performance monitoring and evaluation ensure the stability and reliability of the model throughout the training process, so that the final deployed model can perform well in the actual clinical environment, providing doctors with continuous and accurate auxiliary diagnosis services. In addition, the realization of model reuse and deployment functions allows the trained models to be easily saved, shared and deployed in different medical environments, further expanding the scope of application and influence of the system.

[0059] 2. In the present invention, the image data acquisition module ensures the accurate acquisition of raw data, provides high-quality basic materials for subsequent analysis, and ensures the reliability and effectiveness of diagnosis. The data preprocessing module improves the consistency and availability of data through operations such as denoising, alignment, correction, and normalization, reduces diagnostic errors caused by data quality issues, and lays a solid foundation for feature extraction and model training. The feature extraction module uses deep learning technology to automatically extract key features, greatly improving the efficiency and accuracy of feature extraction, enabling the model to better capture the lesion information in the image, thereby improving the accuracy of diagnosis.

[0060] 3. In the present invention, the diagnosis prediction module uses the trained model to quickly and accurately predict new image data, providing strong decision support for doctors, shortening the diagnosis time, and improving the efficiency of diagnosis and treatment. The result display module displays the diagnosis results in an intuitive and easy-to-understand way, including lesion annotations, prediction probabilities, and detailed diagnosis reports on three-dimensional images, so that doctors and patients can clearly understand the diagnosis, enhance doctor-patient communication, and improve patients' trust and satisfaction. The organic combination of these modules forms an efficient, accurate, and easy-to-use three-dimensional medical imaging diagnosis assistance system, which not only improves the efficiency and quality of medical diagnosis, but also promotes the rapid development of medical imaging diagnosis towards intelligence and precision, bringing revolutionary progress to the medical industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 is the overall system block diagram of the present invention;

[0062] Figure 2 This is a system block diagram of the model training module in the present invention. DETAILED DESCRIPTION

[0063] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0064] Reference Figure 1-2 ,

[0065] A deep learning-driven three-dimensional medical imaging diagnosis auxiliary system, the system includes: an image data acquisition module, a data preprocessing module, a feature extraction module, a model training module, a diagnosis prediction module, and a result display module;

[0066] The model training module is internally configured with a data preparation submodule, a model design submodule, a training process management submodule, a performance evaluation submodule, a hyperparameter adjustment submodule, and a model saving and loading submodule;

[0067] The image data acquisition module is responsible for acquiring the original three-dimensional image data and corresponding annotation information from the medical imaging device, which is then transmitted to the data preprocessing module. The data preprocessing module performs denoising, alignment, correction, normalization, and possible cropping or scaling operations on the original image data, which is converted into a format that meets the model input requirements and stored for subsequent use. The preprocessed data enters the feature extraction module;

[0068] The feature extraction module automatically extracts key features in the image through deep learning techniques such as three-dimensional convolutional neural network (3D CNN). These features will be used to describe the structure and pattern in the image.

[0069] The model training module trains the model based on the features and corresponding annotated data, and adjusts the model parameters through the optimization algorithm to minimize the prediction error. The trained model is used in the diagnosis prediction module to predict new image data and generate diagnosis results.

[0070] The result display module presents the model's prediction results to doctors and patients in an intuitive way, including lesion annotations on three-dimensional images, prediction probabilities, and detailed diagnostic reports. It can also display changes in performance indicators during model training, thus completing the entire process from data collection to result display.

[0071] The image data acquisition module is responsible for acquiring raw three-dimensional image data from various medical imaging devices. The patient's brain is scanned using a magnetic resonance imaging device to generate a series of two-dimensional slices, which are then synthesized into a three-dimensional image. During the data acquisition process, it is necessary to ensure that the resolution, contrast, and signal-to-noise ratio of the image meet the requirements of subsequent processing. In addition, it is necessary to collect patient information corresponding to the image, such as age, gender, clinical diagnosis, etc. This information will be used as annotation data together with the image data for subsequent analysis and model training. After the data acquisition is completed, the image data and annotation data are stored in a secure database for subsequent access and processing.

[0072] In the data preprocessing module, a series of preprocessing operations are performed on the collected raw image data to ensure the quality and consistency of the data and prepare for subsequent feature extraction and model training. First, the three-dimensional image is denoised to reduce the noise introduced by equipment or environmental factors. Next, the image is aligned and corrected to ensure the spatial consistency of images of different patients. In addition, image normalization may be performed to scale the pixel values ​​to a uniform range, such as 0 to 1, to eliminate differences in brightness and contrast between different images. Preprocessing may also include cropping or scaling of the image to match the size requirements of the model input. Finally, data augmentation operations such as random rotation, flipping, or scaling may be performed as needed to increase the generalization ability of the model. The preprocessed image data will be stored and used for subsequent feature extraction and model training.

[0073] The data preparation submodule specifically includes:

[0074] Data loading:

[0075] Load the preprocessed 3D image data X and the corresponding annotation data Y.

[0076] X: A tensor of shape (N,C,D,H,W), where N is the number of samples, C is the number of channels, and D, H, W are depth, height, and width respectively.

[0077] Y: A tensor of shape (N, L), where L is the number of labels.

[0078] Data Augmentation:

[0079] Perform random rotation, scaling, flipping and other operations on X to generate enhanced data X_aug.

[0080] Data segmentation:

[0081] Split X_aug and Y into training set (X_train, Y_train), validation set (X_val, Y_val) and test set (X_test, Y_test).

[0082] The model design submodule specifically includes:

[0083] Network architecture selection:

[0084] Select 3DCNN as the base architecture.

[0085] Layer and parameter configuration:

[0086] Input layer: accepts input of shape (C, D, H, W).

[0087] Convolutional Layer: Multiple 3D convolutional layers are used, and the parameters of each convolutional layer include convolution kernel size k, stride s, and padding p.

[0088] Activation function: ReLU function is usually used.

[0089] Pooling layer: Use a 3D max pooling layer with parameters including pooling kernel size k_pool and stride s_pool.

[0090] Fully connected layer: Flatten the feature map and connect it to the fully connected layer, and finally output the probabilities of L categories.

[0091] Optimizer selection:

[0092] Select Adam optimizer, parameters include learning rate α, weight decay λ, etc.

[0093] The training process management submodule specifically includes:

[0094] Batch processing: X_train and Y_train are processed in batches, with each batch size being B.

[0095] Forward propagation: For each batch of data X_batch, the output Y_pred is calculated through 3DCNN.

[0096] Y_pred = CNN(X_batch)

[0097] Loss calculation

[0098] The loss L is calculated using the cross entropy loss function.

[0099] L=-Σ(Y_true*log(Y_pred)) / B

[0100] Where Y_true is the true label and Y_pred is the probability distribution predicted by the model.

[0101] Back Propagation

[0102] Compute the gradient of the loss function with respect to the network parameters.

[0103] Parameter Update

[0104] The Adam optimizer is used to update the network parameters.

[0105]

[0106] Where θ is the network parameter, α is the learning rate, is the gradient of the loss function with respect to θ.

[0107] The performance evaluation submodules specifically include:

[0108] Indicator calculation: calculate accuracy Acc, recall Rec, and F1 score F1.

[0109] Acc=(TP+TN) / (P+N)

[0110] Rec=TP / P

[0111] F1=2*(Prec*Rec) / (Prec+Rec)

[0112] Where TP is the true positive example, TN is the true negative example, P is the total number of positive examples, N is the total number of negative examples, and Prec is the precision.

[0113] Validation set test: Test the model performance on the validation set and calculate the above indicators.

[0114] Model selection: Select the best model based on the performance indicators on the validation set;

[0115] The hyperparameter adjustment submodule specifically includes:

[0116] Hyperparameter search: Use grid search or random search methods to find the best hyperparameter combination, such as learning rate α, batch size B.

[0117] Cross-validation: Use K-fold cross-validation to evaluate model performance under different hyperparameter combinations;

[0118] The model saving and loading submodule saves the trained model parameters to a file for subsequent use.

[0119] In the feature extraction module, the collected raw image data is preprocessed and feature extracted. First, the three-dimensional image is preprocessed by denoising, alignment, and normalization to reduce the impact of noise and variability. Subsequently, a specific algorithm is used to extract useful features from the preprocessed image. These features may include morphological features (such as volume, shape), texture features (such as gray-level co-occurrence matrix), and intensity features (such as average intensity, standard deviation). The purpose of feature extraction is to convert high-dimensional image data into a lower-dimensional feature representation that is easier for the model to process. The extracted features will be stored and used for subsequent model training.

[0120] In the diagnosis prediction module, the trained model is used to make diagnosis predictions for new image data. When new patient image data is input, the image is first subjected to the same preprocessing and feature extraction operations, and then the extracted features are input into the trained 3D CNN model. The model calculates the output through forward propagation to obtain the probability that the patient belongs to different diagnostic categories. Based on these probabilities, the most likely diagnosis for the patient can be determined. If the model outputs that the probability that the patient has a certain disease exceeds the set threshold (such as 0.5), the disease can be diagnosed. The diagnosis prediction results will be recorded and used for subsequent clinical decision-making.

[0121] In the result display module, the diagnostic prediction results of the model are presented to doctors and patients in an intuitive and easy-to-understand way. A graphical user interface (GUI) can be used to display the patient's three-dimensional image, and the lesion area predicted by the model can be marked on the image. At the same time, a detailed diagnostic report can be provided, including the model's predicted probability, possible disease types, recommended further examinations and treatment plans, etc. In addition, visualization tools (such as TensorBoard) can be used to display changes in loss and performance indicators during model training, so that doctors and researchers can evaluate the performance and stability of the model. The goal of the result display module is to provide comprehensive and accurate information to assist doctors in making better clinical decisions.

[0122] In the present invention, the model training module is the core of the deep learning-driven three-dimensional medical imaging diagnosis auxiliary system, and the careful design and collaborative work of its internal submodules bring significant beneficial effects to the entire system. The data loading submodule ensures the efficient reading and preprocessing of large-scale image data, and provides a stable data stream for subsequent training. Through automated feature learning and model customization and optimization, the model training module realizes the efficient conversion from raw image data to diagnostic knowledge, greatly improving the accuracy and efficiency of diagnosis. Doctors can rely on the accurate diagnosis results provided by the system to formulate treatment plans more quickly and improve patient prognosis. At the same time, efficient training management and performance monitoring and evaluation ensure the stability and reliability of the model throughout the training process, so that the final deployed model can perform well in the actual clinical environment, providing doctors with continuous and accurate auxiliary diagnosis services. In addition, the realization of model reuse and deployment functions allows the trained models to be easily saved, shared and deployed in different medical environments, further expanding the scope of application and influence of the system.

[0123] In the present invention, the image data acquisition module ensures the accurate acquisition of raw data, provides high-quality basic materials for subsequent analysis, and ensures the reliability and effectiveness of diagnosis. The data preprocessing module improves the consistency and availability of data through operations such as denoising, alignment, correction, and normalization, reduces diagnostic errors caused by data quality issues, and lays a solid foundation for feature extraction and model training. The feature extraction module uses deep learning technology to automatically extract key features, greatly improving the efficiency and accuracy of feature extraction, enabling the model to better capture the lesion information in the image, thereby improving the accuracy of diagnosis.

[0124] In the present invention, the diagnosis prediction module uses the trained model to quickly and accurately predict new image data, providing strong decision support for doctors, shortening the diagnosis time, and improving the efficiency of diagnosis and treatment. The result display module displays the diagnosis results in an intuitive and easy-to-understand way, including lesion annotations, prediction probabilities, and detailed diagnosis reports on three-dimensional images, so that doctors and patients can clearly understand the diagnosis, enhance doctor-patient communication, and improve patients' trust and satisfaction. The organic combination of these modules forms an efficient, accurate, and easy-to-use three-dimensional medical imaging diagnosis auxiliary system, which not only improves the efficiency and quality of medical diagnosis, but also promotes the rapid development of medical imaging diagnosis towards intelligence and precision, bringing revolutionary progress to the medical industry.

[0125] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0126] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A deep learning-driven three-dimensional medical imaging diagnostic assistance system, characterized in that: The system includes: an image data acquisition module, a data preprocessing module, a feature extraction module, a model training module, a diagnosis prediction module, and a result display module; The model training module is internally provided with a data preparation submodule, a model design submodule, a training process management submodule, a performance evaluation submodule, a hyperparameter adjustment submodule and a model saving and loading submodule; The image data acquisition module is responsible for acquiring the original three-dimensional image data and corresponding annotation information from the medical imaging device, and these data are then transmitted to the data preprocessing module; the data preprocessing module performs denoising, alignment, correction, normalization, cropping or scaling operations on the original image data, and converts it into a format that meets the model input requirements and stores it for subsequent use; the preprocessed data enters the feature extraction module; The feature extraction module automatically extracts key features in the image through three-dimensional convolutional neural network deep learning technology, and these features will be used to describe the structure and pattern in the image; The model training module features and corresponding annotated data are used to train the model, and the model parameters are adjusted through the optimization algorithm to minimize the prediction error; the trained model is used in the diagnosis prediction module to predict new image data and generate diagnosis results; The result display module presents the prediction results of the model to doctors and patients in an intuitive manner, including lesion annotations on three-dimensional images, prediction probabilities, and detailed diagnostic reports. It also displays changes in performance indicators during model training, thereby completing the entire process from data collection to result display.

2. A deep learning driven three-dimensional medical imaging diagnostic assistance system as claimed in claim 1, characterized in that: The image data acquisition module is responsible for acquiring original three-dimensional image data from various medical imaging devices; using a magnetic resonance imaging device to scan the patient's brain to generate a series of two-dimensional slices, which are then synthesized into a three-dimensional image; During data collection, it is necessary to ensure that the resolution, contrast, and signal-to-noise ratio of the image meet the requirements of subsequent processing; In addition, it is necessary to collect patient information corresponding to the images, including age, gender, and clinical diagnosis. This information will be used as labeled data together with the image data for subsequent analysis and model training. After data collection is completed, the image data and labeled data will be stored in a secure database for subsequent access and processing.

3. The deep learning driven three-dimensional medical imaging diagnosis auxiliary system according to claim 1, characterized in that: The data preprocessing module first performs denoising on the three-dimensional image to reduce the noise introduced by equipment or environmental factors; then, the image is aligned and corrected to ensure the spatial consistency of images of different patients; in addition, the image is normalized to scale the pixel values ​​to a uniform range, such as 0 to 1, to eliminate the differences in brightness and contrast between different images; the preprocessing also includes cropping or scaling the image to match the size requirements of the model input; finally, data enhancement operations are performed as needed, including random rotation, flipping or scaling, to increase the generalization ability of the model.

4. The deep learning driven three-dimensional medical imaging diagnosis auxiliary system according to claim 1, characterized in that: The data preparation submodule specifically includes: Data loading: Load the preprocessed 3D image data X and the corresponding annotation data Y; X: A tensor of shape (N,C,D,H,W), where N is the number of samples, C is the number of channels, and D, H, and W are depth, height, and width, respectively. Y: A tensor of shape (N, L), where L is the number of labels; Data enhancement: Randomly rotate, scale, and flip X to generate enhanced data X_aug; Data splitting: Split X_aug and Y into training set (X_train, Y_train), validation set (X_val, Y_val) and test set (X_test, Y_test).

5. The deep learning driven three-dimensional medical imaging diagnosis auxiliary system according to claim 1, characterized in that: The model design submodule specifically includes: Network architecture selection: Select 3DCNN as the basic architecture; Layer and parameter configuration: Input layer: accepts input of shape (C, D, H, W); Convolutional layer: Use multiple 3D convolutional layers, and the parameters of each convolutional layer include convolution kernel size k, stride s, and padding p; Activation function: Use ReLU function; Pooling layer: Use 3D maximum pooling layer, parameters include pooling kernel size k_pool and step size s_pool; Fully connected layer: Flatten the feature map and connect it to the fully connected layer, and finally output the probabilities of L categories; Optimizer selection: Select the Adam optimizer, and the parameters include learning rate α and weight decay λ.

6. The deep learning driven three-dimensional medical imaging diagnosis auxiliary system according to claim 1, characterized in that: The training process management submodule specifically includes: Batch processing: X_train and Y_train are processed in batches, each batch size is B; Forward propagation: For each batch of data X_batch, the output Y_pred is calculated by 3DCNN; Y_pred = CNN(X_batch) Loss calculation Use the cross entropy loss function to calculate the loss L; L=-Σ(Y_true*log(Y_pred)) / B Where Y_true is the true label and Y_pred is the probability distribution predicted by the model; Back Propagation Calculate the gradient of the loss function with respect to the network parameters; Parameter Update Use Adam optimizer to update network parameters; Where θ is the network parameter, α is the learning rate, is the gradient of the loss function with respect to θ.

7. The deep learning driven three-dimensional medical imaging diagnosis auxiliary system according to claim 1, characterized in that: The performance evaluation submodule specifically includes: Indicator calculation: calculate accuracy Acc, recall Rec, and F1 score F1; Acc=(TP+TN) / (P+N) Rec=TP / P F1=2*(Prec*Rec) / (Prec+Rec) Where TP is the true positive example, TN is the true negative example, P is the total number of positive examples, N is the total number of negative examples, and Prec is the precision; Validation set test: Test the model performance on the validation set and calculate the above indicators; Model selection: Select the best model based on the performance indicators on the validation set; The hyperparameter adjustment submodule specifically includes: Hyperparameter search: Use grid search or random search methods to find the best hyperparameter combination, including learning rate α and batch size B; Cross-validation: Use K-fold cross-validation to evaluate model performance under different hyperparameter combinations; The model saving and loading submodule saves the trained model parameters to a file for subsequent use.

8. The deep learning driven three-dimensional medical imaging diagnosis auxiliary system according to claim 1, characterized in that: In the feature extraction module, the collected raw image data is preprocessed and feature extracted; first, the three-dimensional image is preprocessed by denoising, alignment and normalization to reduce the influence of noise and variability; then, a specific algorithm is used to extract useful features from the preprocessed image; These features include morphological, textural, and intensity features; The purpose of feature extraction is to convert high-dimensional image data into a lower-dimensional feature representation that is easier for the model to process; the extracted features will be stored and used for subsequent model training.

9. The deep learning driven three-dimensional medical imaging diagnosis auxiliary system according to claim 1, characterized in that: In the diagnosis prediction module, the trained model is used to perform diagnosis prediction on new image data; when new patient image data is input, the image is first subjected to the same preprocessing and feature extraction operations, and then the extracted features are input into the trained 3DCNN model; the model calculates the output through forward propagation to obtain the probability that the patient belongs to different diagnostic categories; based on these probabilities, the optimal diagnosis result for the patient is determined; if the model outputs that the probability that the patient has a certain disease exceeds a set threshold, the disease is diagnosed; the diagnosis prediction results will be recorded and used for subsequent clinical decision-making.

10. The deep learning driven three-dimensional medical imaging diagnosis auxiliary system according to claim 1, characterized in that: In the result display module, the diagnostic prediction results of the model are presented to doctors and patients in an intuitive and easy-to-understand manner; a graphical user interface is used to display the patient's three-dimensional image, and the lesion area predicted by the model is marked on the image; at the same time, a detailed diagnosis report is provided, including the model's predicted probability, disease type, and recommended further examinations and treatment plans; in addition, visualization tools are used to display the loss changes and performance indicator changes during the model training process, so that doctors and researchers can evaluate the performance and stability of the model; The goal of the result presentation module is to provide comprehensive and accurate information to assist doctors in making better clinical decisions.

Images