Thoracic surgery image analysis method based on big data

Abstract

The invention discloses a thoracic surgery image analysis method based on big data, and relates to the technical field of thoracic surgery image analysis, and the method comprises the steps: collecting a thoracic surgery image of a patient, carrying out the preprocessing through a standardization protocol, carrying out the automatic recognition of the preprocessed thoracic surgery image through a CNN, and extracting key features, a principal component analysis (PCA) technology is used to carry out dimension reduction processing on extracted key features, representative information is reserved, data of different imaging modes after dimension reduction processing are integrated, a mutual information maximization strategy is used to carry out registration, a support vector machine (SVM) is used as a classification model, and a classification result is obtained. Disease types are classified according to the integrated data of different imaging modes, the quality and efficiency of thoracic surgery image analysis are improved through multi-modal image fusion, feature selection and dimension reduction and model training efficiency, technical innovation in the field of medical image analysis is promoted, and substantial improvement and promotion are brought to clinical practice.

Description

Technical Field

[0001] The present invention relates to the technical field of thoracic surgical image analysis, and particularly to a method for thoracic surgical image analysis based on big data. Background Art

[0002] With the rapid development of medical imaging technology, especially the progress of computed tomography and positron emission tomography imaging modalities, the diagnosis and treatment of thoracic surgical diseases have been improved. Past image analysis mainly relied on the experience of radiologists. To overcome these limitations, automatic image analysis methods based on machine learning algorithms have gradually become a research hotspot. In particular, the application of support vector machines and convolutional neural network technologies enables the extraction of key features from a large amount of image data and classification.

[0003] The current technical solutions are still insufficient in multi-modal image fusion. In particular, there is a lack of effective strategies for integrating data from different imaging modalities, resulting in low information utilization rate and failure to fully utilize their respective advantageous characteristics, thus affecting the final diagnostic accuracy. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a method for thoracic surgical image analysis based on big data to solve the problem of lack of effective strategies for integrating data from different imaging modalities.

[0006] To solve the above technical problems, the present invention provides the following technical solutions: In a first aspect, the present invention provides a method for thoracic surgical image analysis based on big data, which includes: Collecting thoracic surgical images of patients and preprocessing them using a standardized protocol; Automatically identifying and extracting key features from the preprocessed thoracic surgical images using a convolutional neural network (CNN); Using principal component analysis (PCA) technology to perform dimensionality reduction on the extracted key features and retain representative information; Integrating data from different imaging modalities after dimensionality reduction and performing registration using the mutual information maximization strategy; Using a support vector machine (SVM) as a classification model to classify disease types according to the integrated data from different imaging modalities; Based on a real-time optimization feedback mechanism of the classification result, after each classification is completed, adjusting the model parameters according to the classification result and retraining the model using the updated parameters; Combining the re-adjusted model with thoracic surgical images and applying an adaptive learning rate adjustment factor to improve the quality and efficiency of thoracic surgical image analysis.

[0007] As a preferred embodiment of the big data-based thoracic surgery image analysis method of the present invention, wherein: Collect thoracic surgery images of patients and perform preprocessing using a standardized protocol, specifically: Use a CT scanner to obtain thoracic surgery images of patients and use a standardized imaging protocol to set unified scanning parameters; The scanning parameters include slice thickness and scanning range; Perform preprocessing on the thoracic surgery images; The preprocessing includes removing noise and enhancing contrast.

[0008] As a preferred embodiment of the big data-based thoracic surgery image analysis method of the present invention, wherein: Use a convolutional neural network CNN model to identify and extract key features from the preprocessed thoracic surgery images, specifically: Use a ResNet residual network as the architecture of the convolutional neural network CNN model, and input the preprocessed thoracic surgery images into the CNN model for training; Add multiple convolutional layers and pooling layers to the CNN model, and use the convolutional layers to extract key features and use the pooling layers to reduce the spatial dimension of the feature maps; The key features of the thoracic surgery images include shape features, texture features, vascular structures, and tumor boundaries.

[0009] As a preferred embodiment of the big data-based thoracic surgery image analysis method of the present invention, wherein: Use principal component analysis PCA technology to perform dimensionality reduction on the extracted key features and retain representative information, specifically: Perform standardization processing on the key features extracted from the thoracic surgery images; The standardization processing is to subtract the mean value of each key feature from its value and then divide by its standard deviation, so that the mean value of all features is 0 and the standard deviation is 1; Calculate the covariance matrix based on the standardized key feature data, and extract eigenvalues and eigenvectors from the covariance matrix; Select the eigenvector corresponding to the largest eigenvalue as the principal component, project the original thoracic surgery images onto the principal component, and obtain the thoracic surgery images after dimensionality reduction.

[0010] As a preferred embodiment of the big data-based thoracic surgery image analysis method of the present invention, wherein: Use the mutual information maximization strategy to integrate the thoracic surgery images of different imaging modes after dimensionality reduction, and use generative adversarial networks GANs to perform super-resolution reconstruction on the thoracic surgery images, specifically: Integrate the thoracic surgical images of different imaging modalities after dimensionality reduction; The different imaging modalities include CT scan mode and PET scan mode; Use the mutual information maximization strategy to spatially align the images of different imaging modalities for registration, and combine the advantages of different imaging modalities to provide a comprehensive view of the condition; Utilize generative adversarial networks (GANs) to perform super-resolution reconstruction on thoracic surgical images and improve the quality of thoracic surgical images.

[0011] As a preferred embodiment of the big data-based thoracic surgical image analysis method of the present invention, wherein: Use the support vector machine SVM as a classification model to classify the target disease types according to the integrated thoracic surgical images of different imaging modalities, specifically: Use the integrated thoracic surgical images of different imaging modalities as a feature set to train the SVM model; Input the thoracic surgical images into the trained SVM model to classify the disease types.

[0012] As a preferred embodiment of the big data-based thoracic surgical image analysis method of the present invention, wherein: Based on the real-time optimization feedback mechanism of the classification result, after each classification is completed, adjust the model parameters according to the classification result, and use the updated parameters to retrain the model, specifically: Evaluate the classification results of each batch, and calculate the accuracy, precision, and recall of the classification results of this batch; Adjust the penalty coefficient according to the gap between the existing classification results and the target classification results, and use the re-adjusted penalty coefficient to retrain the model; As a preferred embodiment of the big data-based thoracic surgical image analysis method of the present invention, wherein: Use the re-adjusted model combined with thoracic surgical images, and apply an adaptive learning rate adjustment factor to improve the quality and efficiency of thoracic surgical image analysis, specifically: Input the latest thoracic surgical images after preprocessing, feature extraction, and dimensionality reduction into the retrained model, and apply an adaptive learning rate adjustment factor to improve the quality and efficiency of thoracic surgical image analysis.

[0013] In a second aspect, the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and wherein: when the computer program is executed by the processor, any step of the big data-based thoracic surgical image analysis method described in the first aspect of the present invention is implemented.

[0014] In a third aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, wherein: when the computer program is executed by a processor, any step of the method for thoracic surgery image analysis based on big data as described in the first aspect of the present invention is implemented.

[0015] The beneficial effects of the present invention are as follows: through multi-modal image fusion, feature selection and dimensionality reduction, and optimization of the support vector machine model training efficiency, the quality and efficiency of thoracic surgery image analysis are improved. By integrating data from CT and PET imaging modes and adopting the mutual information maximization strategy for registration, a comprehensive view of the condition is provided, and the accuracy of lesion detection is improved. The convolutional neural network CNN is used to automatically extract key features, and dimensionality reduction is performed through the principal component analysis PCA technology to reduce redundant data and enhance the generalization ability of the support vector machine model. The real-time optimization feedback mechanism ensures that the support vector machine model is always in the optimal state, accelerating the convergence speed and performance improvement of the support vector machine model, providing a more accurate and reliable diagnostic tool for clinicians, and helping to detect diseases early and formulate effective treatment plans. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for description in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can also obtain other drawings based on these drawings without creative efforts.

[0017] Figure 1 It is a flowchart of the method for thoracic surgery image analysis based on big data in Embodiment 1.

[0018] Figure 2 It is a schematic diagram of classification optimization feedback in Embodiment 1.

[0019] Figure 3 It is a schematic diagram of feature dimensionality reduction and multi-modal integration in Embodiment 1.

[0020] Figure 4 It is a schematic diagram of data preprocessing and feature extraction in Embodiment 1. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0021] In order to make the above objects, features, and advantages of the present invention more obvious and understandable, the specific embodiments of the present invention will be described in detail below with reference to the drawings in the specification.

[0022] In the following description, numerous specific details are set forth to provide a thorough understanding of the present invention. However, the present invention may be practiced in other ways than those specifically described herein. Those skilled in the art can make similar generalizations without departing from the spirit of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0023] Secondly, as used herein, "an embodiment" or "embodiments" refer to specific features, structures, or characteristics that may be included in at least one implementation of the present invention. The appearances of "in an embodiment" in different places in this specification do not all refer to the same embodiment, nor are they separate or alternative embodiments that exclude each other.

[0024] Example 1, referring to Figures 1 to 4 , this embodiment provides a method for thoracic surgical image analysis based on big data, including the following steps: S1. Collect thoracic surgical images of patients and perform preprocessing using a standardized protocol. Specifically: Select a multi-slice spiral CT scanner with high resolution and low radiation dose, preset the parameters of the multi-slice spiral CT scanner, and set the thickness of each scanned image slice to 1 mm through the touch screen operation interface of the multi-slice spiral CT scanner. A smaller slice thickness can display the scanned image with higher resolution and more detailed anatomical structure information, but it will also increase the scanning time and the amount of thoracic surgical images. For a slice thickness of 1 mm, sufficient thoracic surgical image details can be provided to observe micro-lesions. For chest scans in the area from above the collarbone to the lower edge of the ribs, guide the patient to take a supine position, raise both arms, and keep still to avoid motion artifacts affecting the quality of thoracic surgical images. Also, it is necessary to cover the entire lungs and heart to ensure that no anatomical structures and lesions are missed. When observing the vascular structures in the chest, such as the aorta, pulmonary artery, and veins, start scanning about 30 seconds after injecting an iodine-based contrast agent at the patient's blood vessels to optimally display the vascular structures Start the multi-slice spiral CT scanner according to the above method, complete the tomographic scan within the range from above the collarbone to the lower edge of the ribs, obtain thoracic surgical images, and perform denoising and contrast enhancement.

[0025] Apply the non-local means filtering algorithm. This algorithm reduces noise by comparing the similarity between different pixel blocks in the thoracic surgical images obtained by the multi-slice spiral CT scanner, expressed as ; where is the pixel value after filtering, is the target pixel position, is the reference pixel position, is the weight function; The gray-scale mapping is calculated using the cumulative distribution function (CDF) to enhance the overall contrast of the image, which is expressed as, ; where, represents the gray-scale mapping, is the cumulative distribution function (CDF), is the maximum gray level of the thoracic surgical image, usually 255, represents the gray level; The thoracic surgical images that have been denoised and contrast-enhanced are saved in the DICOM standard format, which helps to maintain the consistency and traceability of the data.

[0026] S2. Use the convolutional neural network (CNN) model to identify and extract key features from the preprocessed thoracic surgical images. Specifically, Select the ResNet50 residual network architecture as the basic architecture of the convolutional neural network (CNN). The ResNet50 residual network architecture consists of multiple residual blocks, and each residual block contains several convolutional layers and a skip connection, which can effectively alleviate the problem of gradient disappearance in deep networks.

[0027] The thoracic surgical images that have been denoised and contrast-enhanced are input into the convolutional neural network (CNN) model in batches of 32. Each image is first scaled to a unified size to ensure the consistency of the training of the convolutional neural network (CCN) model.

[0028] Based on the ResNet50 residual network architecture, additional convolutional layers are added to further refine the extraction of key features. A convolutional layer with 64 filter kernels of size 3x3 is added before each residual block. To reduce the spatial dimension of the feature map and control overfitting, a max pooling layer is added after each residual block, with the pooling window size set to 2x2 and the stride set to 2. The above steps can help extract important key features, including shape features, texture features, vascular structures, and tumor boundaries, while also reducing the computational complexity; Shape features refer to the geometric properties of objects in the thoracic surgical image, such as edges and contours, which help to identify the contours of organs, such as the lungs and heart, and observe whether there are abnormal hyperplasia and lesions in the thoracic surgical image; Texture features describe the repeated patterns and local variations within the thoracic surgical image region, reflecting the microscopic structure inside the tissue, and are used to identify the lesion regions of lung nodules and tumors. For example, benign tumors and malignant tumors may exhibit different texture patterns, which helps doctors make more accurate diagnoses; The vascular structure refers to the distribution and connection pattern of blood vessels shown in the image. The vascular structure is extremely important for surgical planning. For example, during lung cancer resection, clearly understanding the location and distribution of blood vessels around the tumor can help surgeons design a safer and more effective surgical plan and reduce the risk of bleeding; The tumor boundary refers to the boundary between the tumor and the surrounding normal tissues. Accurately defining the tumor boundary is crucial for assessing the tumor size, location, and formulating treatment strategies. In addition, it can also be used to monitor the treatment effect, such as observing the change in tumor volume after radiotherapy; The extraction and mutual complementation of shape features, texture features, vascular structure, and tumor boundary constitute the basis for the interpretation of thoracic surgical images.

[0029] S3. Use the principal component analysis (PCA) technique to perform dimensionality reduction on the extracted key features. Specifically, Extract the key features from the convolutional layer of the CNN model above, including shape features, texture features, vascular structure, and tumor boundary. Calculate the mean and standard deviation for each key feature, expressed as, ; ; Among them, represents the mean of the eigenvalue set, represents the number of samples of the eigenvalue, represents the th sample's eigenvalue, represents the standard deviation of the eigenvalue set; Apply a standardization operation to each feature so that all feature means are 0 and the standard deviation is 1. The standardized eigenvalue is expressed as, ; Among them, represents the standardized eigenvalue; Arrange all the standardized eigenvalues in columns to form an n×m matrix, where n represents the number of samples and m represents the number of features.

[0030] Based on the standardized feature matrix, calculate the covariance matrix, expressed as, ; Among them, C represents the covariance matrix, represents transpose matrix, represents the matrix of the combined standardized eigenvalues; Perform eigenvalue decomposition on the covariance matrix. The goal is to find the eigenvalues and eigenvectors of the covariance matrix, expressed as, ; Among them, represents the eigenvalues of the matrix, represents the eigenvectors; Obtain the eigenvalues and their corresponding eigenvectors. Sort them according to the eigenvalue magnitudes and select the eigenvector corresponding to the largest eigenvalue as the principal component. The selection of eigenvalues can be determined based on the cumulative contribution rate. For example, select the eigenvalues with a cumulative contribution rate reaching 80% - 95% as the principal components.

[0031] Project the standardized feature matrix onto the principal components to obtain the dimensionality-reduced thoracic surgery images.

[0032] S4. Use the mutual information maximization strategy to integrate the dimensionality-reduced thoracic surgery images of different imaging modalities, and utilize the generative adversarial network GANs to perform super-resolution reconstruction on the thoracic surgery images. Specifically, For the thoracic surgery images of CT scans and PET scans after PCA dimensionality reduction obtained above, ensure that the data of each imaging modality have been standardized and have the same number of samples and corresponding label information. Adopt the mutual information maximization strategy to find the optimal spatial transformation parameters by maximizing the mutual information between two images for spatial alignment. Specifically, through spatial transformation parameters such as translation, rotation, and scaling. This is achieved by setting a reasonable range, for example, the translation range is within ±10 pixels and the rotation angle is within ±5 degrees. For each pair of CT images and PET images to be registered, use image processing software such as ITK-SNAP and SimpleITK to perform mutual information under different spatial transformations. Mutual information measures the amount of information shared between two images. Perform spatial alignment through the spatial transformation of mutual information executed by the image processing software.

[0033] The pixel value of each pixel in the CT image represents the absorption amount of X-rays at that pixel point, usually expressed in Hounsfield units (HU). For example, the absorption amount of water is 0 HU, and that of bone may exceed +1000 HU. The pixel value of each pixel in the PET image reflects the concentration of the radioactive tracer at that pixel point, which is usually related to metabolic activity. The pixel values of the PET image are usually quantified by the standardized uptake value (SUV).

[0034] Directly add the CT pixel value and the PET pixel value at the same spatial position, and use the calculated new value as the pixel value at the corresponding position in the new image. Similarly, add all the CT pixel values and PET pixel values at the corresponding positions through the above method and combine them all to form a new fused image. Considering the advantages of different imaging modalities, the eigenvectors of the CT and PET thoracic surgery images can also be concatenated to form a new comprehensive eigenvector.

[0035] Using a medical image viewer, such as 3D Slicer, to load the fused thoracic surgical images can provide a more comprehensive view of the condition than a single imaging modality. For example, CT images provide detailed anatomical information, while PET images show metabolic activity. Combining the two can help doctors better understand the physiological state of the lesion area and gain a more comprehensive understanding of the condition. When the quality of thoracic surgical images is found to be unsatisfactory (such as low resolution, high noise, etc.), the GANs enhancement process is initiated. The trained GANs model is used to perform super-resolution reconstruction on the selected thoracic surgical images. Specifically, for each thoracic surgical image to be enhanced, the generator network first generates a high-resolution version and a new perspective image, and then the discriminator network evaluates the authenticity of the thoracic surgical image. This process is iterated repeatedly until a satisfactory result is obtained. Realistic medical images are generated through the GANs model to improve low-quality and partially missing image data.

[0036] In addition to CT and PET scan modes, other imaging modalities, such as MRI and ultrasound, can also be considered for integration to form a more comprehensive multi-modal image dataset. Different imaging modalities provide complementary information. The thoracic surgical images obtained by combining multiple imaging modalities can provide doctors with a more three-dimensional and comprehensive view of the condition, helping to more accurately judge the disease state and formulate treatment plans.

[0037] S5. Use the support vector machine (SVM) as a classification model to classify the target disease types based on the fused thoracic surgical images of different imaging modalities. Specifically, Based on the good performance of the support vector machine in high-dimensional space, the support vector machine with a linear kernel is selected as the classification model. Moreover, the form of the linear kernel function in the support vector machine model is simple and the computational efficiency is high, which is suitable for medical image analysis.

[0038] The cross-validation method is used to determine the optimal penalty coefficient. Specifically, in a 5-fold cross-validation process, a series of penalty coefficients, such as 0.01, 0.1, 1, 10, 100, are tried, and the penalty coefficient that maximizes the accuracy of the validation set is selected.

[0039] The fused comprehensive feature vectors are randomly divided into a training set and a test set and divided according to the ratio of 80% training set and 20% test set to ensure that the samples of each target disease type are reasonably distributed in the training set and the test set. The training set is input into the support vector machine model for training.

[0040] After preprocessing, dimensionality reduction, and fusion steps, the thoracic surgical images are used as samples and input into a pre-trained support vector machine model. The support vector machine model will determine which class, positive or negative, the new sample belongs to according to the rules learned internally. Specifically, the support vector machine model calculates a score for each new sample and then determines which class the sample belongs to based on this score. For example, if the score is greater than a certain threshold, the sample is considered to belong to the positive class; otherwise, it belongs to the negative class. If the prediction result of the support vector machine model is the positive class, it indicates that the disease type predicted by the support vector machine model for this sample is consistent with the target disease type; if it is the negative class, it means it is inconsistent with the target disease type. This explanation can help doctors understand the patient's condition or pathological changes. If necessary, the prediction result can also be marked on the original thoracic surgical image to help doctors better understand the decision-making basis of the support vector machine model. For example, the lesion area can be marked in the image and the predicted classification result can be displayed, which can be achieved through professional medical image viewing software.

[0041] S6. Real-time optimization feedback mechanism based on the classification result. After each classification is completed, the classification model parameters are adjusted according to the classification result, and the updated parameters are used to retrain the classification model. Specifically, According to the prediction of the support vector machine model, if the prediction result of the support vector machine model is the positive class, it indicates that the disease type predicted by the support vector machine model for this sample is consistent with the target disease type; if it is the negative class, it means it is inconsistent with the target disease type. The number of all positive-class samples divided by the total number of samples represents the accuracy rate. Specifically, it checks how many samples have a predicted disease type consistent with the target disease type. Among all the samples predicted to be positive, such as the lesion areas, the proportion that is actually the positive class is represented as the precision rate. Among all the samples that are actually positive, the number that is correctly predicted as positive is represented as the recall rate.

[0042] Compare the performance metrics of the current support vector machine model, including accuracy rate, precision rate, and recall rate, with the ideal performance metrics. The ideal performance metrics are set based on domain knowledge and experimental results. According to the difference between the current support vector machine model's performance metrics and the ideal performance metrics, adjust the penalty coefficient in the support vector machine model. For example, if the precision rate of the support vector machine model is too low, the penalty coefficient can be increased to reduce false alarms. On the contrary, if the recall rate is low, the penalty coefficient needs to be decreased to increase the recall rate. Here, we adjust the penalty coefficient in the support vector machine model according to the actual situation in life and the above method of adjusting the penalty coefficient, and use the adjusted penalty coefficient and the training set to input into the support vector machine model for training.

[0043] S7. Use the readjusted model combined with thoracic surgical images and apply an adaptive learning rate adjustment factor to improve the quality and efficiency of thoracic surgical image analysis. Specifically, The new thoracic surgical imaging data is preprocessed and feature-extracted according to the steps mentioned above, and a new comprehensive feature vector is formed by connecting the feature vectors of the CT and PET thoracic surgical images. The retrained support vector machine model is used to predict the new samples. For each new sample, if the score is greater than 0, the sample is predicted to belong to the positive class; if it is less than 0, it belongs to the negative class.

[0044] By retraining and applying the support vector machine model and introducing an adaptive learning rate adjustment factor at the same time, the purpose of designing this factor is to improve the generalization ability of the support vector machine model by increasing the learning rate of samples close to the decision boundary, and the adaptive learning rate adjustment factor continuously self-adjusts without interrupting the workflow, improving the learning efficiency and classification accuracy of the support vector machine model. By comparing the performance of the old and new support vector machine models in thoracic surgical image analysis, the improvements of the support vector machine model are recorded, and a feedback mechanism is established to collect the feedback from doctors and patients. The support vector machine model is adjusted accordingly based on the feedback from doctors and patients. These feedbacks can help identify potential problems of the support vector machine model and guide future improvement directions, providing more accurate diagnostic information for clinicians and helping to develop more effective personalized treatment plans.

[0045] This embodiment also provides a computer device applicable to the case of the thoracic surgical imaging analysis method based on big data, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the thoracic surgical imaging analysis method based on big data proposed in the above embodiment.

[0046] This computer device can be a terminal. The computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. Among them, the processor of this computer device is used to provide computing and control capabilities. The memory of this computer device includes a non-volatile storage medium and an 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 the computer program in the non-volatile storage medium. The communication interface of this computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, a carrier network, NFC (Near Field Communication) or other technologies. The display screen of this computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of this computer device can be a touch layer covered on the display screen, or a button, a trackball or a touchpad set on the housing of the computer device, or an external keyboard, a touchpad or a mouse, etc.

[0047] This embodiment also provides a storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the method for thoracic surgery image analysis based on big data proposed in the above embodiment. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM for short), electrically erasable programmable read-only memory (EEPROM for short), erasable programmable read-only memory (EPROM for short), programmable read-only memory (PROM for short), read-only memory (ROM for short), magnetic memory, flash memory, magnetic disk or optical disc.

[0048] In summary, the present invention improves the quality and efficiency of thoracic surgery image analysis through multi-modal image fusion, feature selection and dimensionality reduction, and optimization of the support vector machine model training efficiency. By integrating the data of CT and PET imaging modes and adopting the mutual information maximization strategy for registration, a comprehensive view of the condition is provided, improving the accuracy of lesion detection. The convolutional neural network CNN is used to automatically extract key features, and dimensionality reduction is performed through the principal component analysis PCA technology, reducing redundant data and enhancing the generalization ability of the support vector machine model. The real-time optimization feedback mechanism ensures that the support vector machine model is always in the optimal state, accelerating the convergence speed and performance improvement of the support vector machine model, providing a more accurate and reliable diagnostic tool for clinicians, and helping to detect diseases early and formulate effective treatment plans.

[0049] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A method for thoracic surgical image analysis based on big data, characterized in that: including, collecting thoracic surgical images of patients and preprocessing them using a standardized protocol; using a Convolutional Neural Network (CNN) to identify and extract key features from the preprocessed thoracic surgical images; using Principal Component Analysis (PCA) technology to reduce the dimensionality of the extracted key features; using the mutual information maximization strategy to integrate thoracic surgical images of different imaging modalities after dimensionality reduction, and using Generative Adversarial Networks (GANs) to perform super-resolution reconstruction on the thoracic surgical images; using Support Vector Machine (SVM) as a classification model to classify target disease types based on the integrated thoracic surgical images of different imaging modalities; based on a real-time optimization feedback mechanism for classification results, after each classification is completed, adjusting the classification model parameters according to the classification results, and retraining the classification model using the updated parameters; combining the re-adjusted classification model with thoracic surgical images, and applying an adaptive learning rate adjustment factor to improve the quality and efficiency of thoracic surgical image analysis.

2. The method for thoracic surgical image analysis based on big data according to claim 1, wherein: Collecting thoracic surgical images of patients and preprocessing them using a standardized protocol, specifically: using a CT scanner to obtain thoracic surgical images of patients and using a standardized imaging protocol to set unified scanning parameters; the scanning parameters include slice thickness and scanning range; preprocessing the thoracic surgical images; the preprocessing includes removing noise and enhancing contrast.

3. The method for thoracic surgical image analysis based on big data according to claim 2, characterized in that: Using a Convolutional Neural Network (CNN) model to identify and extract key features from the preprocessed thoracic surgical images, specifically: using a ResNet residual network as the architecture of the Convolutional Neural Network (CNN) model, and inputting the preprocessed thoracic surgical images into the CNN model for training; adding multiple convolutional layers and pooling layers to the CNN model, and using convolutional layers to extract key features and using pooling layers to reduce the spatial dimension of the feature maps; the key features of the thoracic surgical images include shape features, texture features, vascular structure, and tumor boundary.

4. The method for thoracic surgical image analysis based on big data according to claim 3, characterized in that: Using Principal Component Analysis (PCA) technology to reduce the dimensionality of the extracted key features, specifically: standardizing the key features extracted from the thoracic surgical images; the standardization process is to subtract the mean value of each key feature from its value and then divide by its standard deviation, so that all feature means are 0 and the standard deviation is 1; calculating the covariance matrix based on the standardized key feature data, and extracting eigenvalues and eigenvectors from the covariance matrix; selecting the eigenvector corresponding to the largest eigenvalue as the principal component, and projecting the original thoracic surgical images onto the principal component to obtain the thoracic surgical images after dimensionality reduction.

5. The method for thoracic surgical image analysis based on big data according to claim 4, wherein: Using the mutual information maximization strategy to integrate thoracic surgical images of different imaging modalities after dimensionality reduction, and using Generative Adversarial Networks (GANs) to perform super-resolution reconstruction on the thoracic surgical images, specifically: integrating the thoracic surgical images of different imaging modalities after dimensionality reduction; the different imaging modalities include CT scan mode and PET scan mode; using the mutual information maximization strategy to register different imaging modality images by aligning them spatially, combining the advantages of different imaging modalities to provide a comprehensive view of the condition; using Generative Adversarial Networks (GANs) to perform super-resolution reconstruction on the thoracic surgical images and improve the quality of the thoracic surgical images.

6. The method for thoracic surgery image analysis based on big data according to claim 5, characterized in that: Use the support vector machine (SVM) as the classification model to classify disease types according to the integrated thoracic surgical images of different imaging modalities, specifically as follows: Use the integrated thoracic surgical images of different imaging modalities as the feature set to train the SVM model; Input the thoracic surgical images into the trained SVM model to classify disease types.

7. The method for thoracic surgical image analysis based on big data according to claim 6, characterized in that: Based on the real-time optimization feedback mechanism of the classification results, after each classification is completed, adjust the classification model parameters according to the classification results, and use the updated parameters to retrain the classification model, specifically as follows: Evaluate the classification results of each batch, and calculate the accuracy, precision, and recall rate of the classification results of this batch; Adjust the penalty coefficient according to the gap between the existing classification results and the target classification results, and use the re-adjusted penalty coefficient to retrain the model.

8. The method for analyzing thoracic surgical images based on big data according to claim 7, wherein: Use the re-adjusted model combined with thoracic surgical images, and apply the adaptive learning rate adjustment factor to improve the quality and efficiency of thoracic surgical image analysis, specifically as follows: Input the latest thoracic surgical images after preprocessing, feature extraction, and dimensionality reduction into the retrained classification model, and apply the adaptive learning rate adjustment factor to improve the quality and efficiency of thoracic surgical image analysis.

9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the big data-based thoracic surgical image analysis 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 the processor, it implements the steps of the big data-based thoracic surgical image analysis method according to any one of claims 1 to 8.

Images