A method, medium and system for detecting biological activity in a tumor section preservation process

Abstract

The application provides a tumor slice preservation process biological activity detection method, medium and system, and belongs to the tumor slice preservation technical field, and comprises the following steps: first, a high-definition image and a plurality of key point microscopic images of the slice just placed in a preservation device are acquired to form a first group of image data sets. Then, a high-definition image and a key point microscopic image at a detection time are acquired to form a second group of image data sets. The two groups of high-definition images and microscopic images are aligned and feature-extracted, input into a pre-trained biological activity detection model, and finally, a current activity vector of the tumor slice is output. The method utilizes computer vision technology to track the subtle changes of the slice in the preservation process, and can provide valuable information for cancer diagnosis and treatment. The method, medium and system solve the technical problem that the existing image analysis-based method cannot comprehensively reflect the multi-level structure and functional changes of the tumor slice in the preservation process.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of tumor section preservation, and particularly relates to a tumor section preservation process biological activity detection method, medium and system. BACKGROUND

[0002] In recent years, tumor section specimens play an increasingly important role in the fields of pathological diagnosis, drug screening and the like. These tumor section samples usually need to be preserved for a long time under specific preservation conditions to ensure that the biological activity thereof does not change significantly in subsequent detection and analysis processes. Existing common tumor section preservation methods include low-temperature preservation, chemical fixation and the like, but these methods often cause different degrees of influence on the microstructure and physiological function of the section. Therefore, how to accurately evaluate the biological activity change of the tumor section in the preservation process has become one of the key problems to be solved in the field.

[0003] At present, some studies have attempted to use high-definition imaging and microscopic imaging technology to quantitatively analyze the structural and functional characteristics of tumor sections, and evaluate the biological activity change thereof in the preservation process. For example, some scholars propose to analyze the texture features and shape features of the high-definition image of the tumor section to reflect the change of the overall structure of the section. Some studies use microscopic imaging technology to collect dynamic change data of the microcellular structure of the section to reveal the evolution law of the functional characteristics thereof. However, these methods usually only focus on single-scale features, and it is difficult to comprehensively reflect the multi-level structural and functional changes of the section in the preservation process. In addition, due to the great difference between different section samples in the initial state and in the preservation process, a single feature analysis method often cannot accurately predict the current biological activity level. SUMMARY

[0004] Therefore, the application provides a tumor section preservation process biological activity detection method, medium and system, which can solve the technical problem that the existing image analysis-based method cannot comprehensively reflect the multi-level structural and functional changes of the tumor section in the preservation process.

[0005] The application is implemented as follows:

[0006] A first aspect of the application provides a tumor section preservation process biological activity detection method, which comprises the following steps:

[0007] S10, obtaining a high-definition image of a tumor section just placed in a preservation device, and obtaining a first high-definition image after preprocessing

[0008] S20, obtaining a plurality of key points by using a scale-invariant feature transformation method according to the first high-definition image, and obtaining a microscopic image of each key point to form a first microscopic image set comprising a plurality of first microscopic images;

[0009] S30, obtain a high-definition image of the tumor section to be detected, obtain a second high-definition image after preprocessing, and obtain a corresponding microscopic image according to the plurality of key points to form a second microscopic image set including a plurality of second microscopic images;

[0010] S40, align the first high-definition image and the second high-definition image, and align the first microscopic image and the second microscopic image of each key point;

[0011] S50, respectively extract features of the aligned first high-definition image and the second high-definition image, denoted as first high-definition features and second high-definition features; and respectively extract features of the first microscopic image and the second microscopic image of each key point, denoted as first microscopic features and second microscopic features;

[0012] S60, input the first high-definition features and the second high-definition features, the first microscopic features and the second microscopic features of each key point into a pre-trained tumor section preservation process biological activity detection model to obtain a current activity vector of the tumor section and output.

[0013] The tumor section preservation process biological activity detection model adopts a multi-scale feature fusion structure, including a high-definition feature subnetwork, a microscopic macroscopic feature subnetwork, a microscopic microscopic feature subnetwork, a microscopic texture feature subnetwork, and a fusion subnetwork.

[0014] Further, the high-definition feature subnetwork is used to extract the overall features of the high-definition image, the input is the first high-definition feature and the second high-definition feature, and the output is the high-definition image feature change vector; and the structure is a convolutional neural network based on ResNet50.

[0015] Further, the microscopic macroscopic feature subnetwork is used to extract the macroscopic structure features of the microscopic image, the input is the first microscopic feature and the second microscopic feature, and the output is the microscopic macroscopic feature change vector; and the structure is a convolutional neural network based on VGG16.

[0016] Further, the microscopic microscopic feature subnetwork is used to extract the microscopic cell features of the microscopic image, the input is the first microscopic feature and the second microscopic feature, and the output is the microscopic microscopic feature change vector; and the structure is a convolutional neural network based on DenseNet121.

[0017] Further, the microscopic texture feature subnetwork is used to extract the texture features of the microscopic image, the input is the first microscopic feature and the second microscopic feature, and the output is the microscopic texture feature change vector; and the structure is a feature extraction network based on a gray level co-occurrence matrix and a local binary pattern.

[0018] Further, the fusion sub-network is used for fusing the features extracted by each sub-network, and the input is a high-definition image feature change vector, a microscopic macroscopic feature change vector, a microscopic microscopic feature change vector and a microscopic texture feature change vector, and the output is a current activity vector of the tumor section; and the structure is a multi-layer perception network.

[0019] The following is the training data set establishment step of the biological activity detection model of the tumor section preservation process:

[0020] 1. Sample collection: collect tumor section samples of different types and different stages, and take high-definition images and microscopic images of each sample at different preservation time points (such as 0h, 2h, 4h, 6h……72h, etc.);

[0021] 2. Data preprocessing: standardizing, denoising and other preprocessing of high-definition images, extracting key points using SIFT algorithm, and obtaining microscopic images of key points;

[0022] 3. Feature extraction: extracting high-definition image features, extracting macroscopic structure features, microscopic cell features and texture features of microscopic images;

[0023] 4. Activity annotation: using experimental method to evaluate the activity of each time point section sample, and converting the evaluation result into a numerical activity vector;

[0024] 5. Data augmentation: rotating, flipping, scaling and other operations are performed on the images to expand the data set, and synthetic samples are generated using GAN and other technologies to further increase the diversity of data;

[0025] 6. Data set division: dividing the processed data into training set, validation set and test set according to the ratio of 7:2:1.

[0026] Further, the preprocessing includes image standardization, image enhancement, and denoising processing.

[0027] Specifically, the step S10 specifically includes:

[0028] Step 101, collecting tumor section samples and placing them in special preservation instruments. High-definition images of the tumor section are immediately taken using a high-definition camera device to obtain a first high-definition image.

[0029] Step 102, performing image standardization, image enhancement, denoising processing and other preprocessing operations on the first high-definition image to improve the quality and usability of the image, and obtaining a final first high-definition image.

[0030] Specifically, the step S20 specifically includes:

[0031] Step 201, using a scale-invariant feature transform (SIFT) algorithm to detect feature points of the first high-definition image, and obtaining a plurality of significant key points.

[0032] Step 202, for each key point, using a high-power microscopic imaging device to capture a local microscopic image of the tumor section tissue at the position, to form a first microscopic image set.

[0033] The step S30 specifically comprises:

[0034] Step 301, when it is necessary to detect the biological activity change of the tumor section in the preservation process, using a high-definition camera device to capture a high-definition image of the section again, to obtain a second high-definition image.

[0035] Step 302, using a SIFT algorithm to detect key points of the second high-definition image, and using a microscopic imaging device to obtain local microscopic images at the key points, to form a second microscopic image set.

[0036] The step S40 specifically comprises:

[0037] Step 401, using a RANSAC image registration algorithm based on feature points to perform geometric correction and registration on the first high-definition image and the second high-definition image, to eliminate the difference in shooting, so that the two high-definition images are aligned in spatial position.

[0038] Step 402, for each key point, its corresponding first microscopic image and second microscopic image are also locally registered based on SIFT features, to ensure that the microscopic images of the same key point shot at two time points are completely aligned in spatial position.

[0039] The step S50 specifically comprises:

[0040] Step 501, using a pre-trained ResNet50 convolutional neural network to extract the overall features of the first high-definition image and the second high-definition image respectively, denoted as first high-definition features and second high-definition features.

[0041] Step 502, extracting the macrostructure features (based on VGG16), microscopic cell features (based on DenseNet121) and texture features (based on GLCM and LBP) of the first microscopic image and the second microscopic image corresponding to each key point respectively, denoted as first microscopic features and second microscopic features.

[0042] The step S60 specifically comprises:

[0043] Step 601, the tumor section preservation process biological activity detection model adopts a multi-scale feature fusion structure, including a high-definition feature subnetwork, a microscopic macro feature subnetwork, a microscopic micro feature subnetwork, a microscopic texture feature subnetwork, and a fusion subnetwork.

[0044] Step 602, the high-definition feature subnetwork is based on a ResNet50 structure and is used for extracting overall features of a high-definition image, the input is a first high-definition feature and a second high-definition feature, and the output is a high-definition image feature change vector.

[0045] Step 603, the microscopic macro feature subnetwork is based on a VGG16 structure and is used for extracting macro structure features of a microscopic image, the input is a first microscopic feature and a second microscopic feature, and the output is a microscopic macro feature change vector.

[0046] Step 604, the microscopic micro feature subnetwork is based on a DenseNet121 structure and is used for extracting micro cell features of a microscopic image, the input is a first microscopic feature and a second microscopic feature, and the output is a microscopic micro feature change vector.

[0047] Step 605, the microscopic texture feature subnetwork is based on GLCM and LBP and is used for extracting texture features of a microscopic image, the input is a first microscopic feature and a second microscopic feature, and the output is a microscopic texture feature change vector.

[0048] Step 606, the fusion subnetwork adopts an MLP structure and is used for fusing features extracted by each subnetwork, the input is a high-definition image feature change vector, a microscopic macro feature change vector, a microscopic micro feature change vector, and a microscopic texture feature change vector, and the output is a current activity vector of a tumor section.

[0049] Optionally, the high-definition camera equipment used in the step 101 includes a digital camera or a microscopic imaging system, and the image preprocessing operation in the step 102 includes image standardization, histogram equalization, edge sharpening, median filtering, or bilateral filtering.

[0050] Optionally, the SIFT algorithm adopted in the step 201 can stably detect key points under conditions of image scale, rotation, and illumination changes.

[0051] Optionally, the high-definition camera equipment used in the step 301 is of the same type as the high-definition camera equipment used in the step 101.

[0052] Optionally, the RANSAC image registration algorithm adopted in the step 401 can eliminate differences in translation, rotation, scaling, and the like when shooting, so that two high-definition images are accurately aligned in spatial position.

[0053] Optionally, the microscopic image features extracted in the step 502 include macrostructure features, microscopic cell features and texture features, which can comprehensively reflect the multi-scale information of the slice tissue.

[0054] Optionally, the output feature change vectors of the high-definition feature sub-network, the microscopic macro-feature sub-network, the microscopic micro-feature sub-network and the microscopic texture feature sub-network can comprehensively reflect the multi-scale structure and texture change features of the tumor slice in the preservation process.

[0055] Optionally, the fusion sub-network adopts a multi-layer perception (MLP) structure, which can effectively fuse the features extracted by each sub-network and output the current biological activity vector of the tumor slice.

[0056] Optionally, the steps S10 to S60 constitute a complete tumor slice preservation process biological activity detection process, which can comprehensively evaluate the structure and function changes of the slice in the preservation process.

[0057] Optionally, the method can be applied to tumor slice samples of different types and different stages to detect and monitor the biological activity changes thereof in the preservation process.

[0058] Optionally, the training data collected by the method includes high-definition images and microscopic images of tumor slices of different types and at different preservation time points, and the activity at each time point is experimentally evaluated and labeled, thereby establishing a representative training data set; the training data set is subjected to data enhancement operations such as rotation, flipping and scaling, and synthetic samples are generated by using techniques such as a generative adversarial network (GAN), thereby further increasing the diversity and generalization ability of the data.

[0059] The second aspect of the application provides a computer readable storage medium, wherein the computer readable storage medium stores program instructions, and the program instructions are used to execute the tumor slice preservation process biological activity detection method.

[0060] The third aspect of the application provides a tumor slice preservation process biological activity detection system, which comprises the computer readable storage medium.

[0061] Compared with the prior art, the tumor slice preservation process biological activity detection method, medium and system have the following advantages: 1) the method fully utilizes the information of high-definition images and microscopic images of two different scales, can comprehensively reflect the multi-level structure and function change features of the slice in the preservation process, and is more accurate and reliable than single-scale feature analysis.

[0062] 2) The method uses key point detection and local microscopic image acquisition technology based on SIFT algorithm, which can accurately align high-definition images and microscopic images taken at different time points, laying a foundation for subsequent feature comparison and analysis.

[0063] 3) The method constructs a multi-scale feature fusion model including a high-definition feature subnetwork, a microscopic macro feature subnetwork, a microscopic micro feature subnetwork, and a microscopic texture feature subnetwork, which can effectively extract and fuse various features and output the current biological activity level of the slice, providing a reliable basis for slice preservation and management.

[0064] 4) The method collects rich tumor slice sample data, including high-definition images and microscopic images of different types and different preservation time points, as well as corresponding experimental activity evaluation results, providing sufficient support for model training. At the same time, the generalization ability of the model is further improved through data enhancement and other means.

[0065] In summary, the tumor slice preservation process biological activity detection method proposed in the present application can comprehensively evaluate the multi-scale structure and functional changes of the slice during the preservation process, and significantly improves the accuracy and reliability compared to the prior art. It solves the technical problem of the existing image analysis-based method that it is difficult to comprehensively reflect the multi-level structure and functional changes of the tumor slice during the preservation process. BRIEF DESCRIPTION OF DRAWINGS

[0066] Figure 1 The flowchart of the method provided by the present application is shown. DETAILED DESCRIPTION

[0067] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be described clearly and completely with reference to the drawings in the embodiments of the present application.

[0068] As shown in Figure 1 , it is a tumor slice preservation process biological activity detection method flowchart provided by the present application, and the method includes the following steps:

[0069] S10, obtaining a high-definition image of a tumor slice just placed in a preservation device, and obtaining a first high-definition image after preprocessing

[0070] S20, obtaining a plurality of key points by using a scale-invariant feature transformation method according to the first high-definition image, and obtaining a microscopic image of each key point to form a first microscopic image set including a plurality of first microscopic images;

[0071] S30, obtain a high-definition image of the tumor section at the detection time, obtain a second high-definition image after preprocessing, and obtain a corresponding microscopic image according to the plurality of key points to form a second microscopic image set including a plurality of second microscopic images;

[0072] S40, align the first high-definition image and the second high-definition image, and align the first microscopic image and the second microscopic image of each key point;

[0073] S50, extract features of the aligned first high-definition image and the second high-definition image, denoted as first high-definition features and second high-definition features; and extract features of the first microscopic image and the second microscopic image of each key point, denoted as first microscopic features and second microscopic features;

[0074] S60, input the first high-definition features and the second high-definition features, and the first microscopic features and the second microscopic features of each key point into a pre-trained tumor section preservation process biological activity detection model to obtain a current activity vector of the tumor section and output.

[0075] The specific implementation of the above steps is described in detail as follows:

[0076] Step S10: obtain a high-definition image of the tumor section immediately after being placed in the preservation device, and obtain a first high-definition image after preprocessing

[0077] The specific implementation is as follows:

[0078] 1) Collect the tumor section sample and place it in a dedicated preservation device. Use a high-definition camera device (such as a high-resolution digital camera or a microscopic imaging system) to immediately take a high-definition image of the tumor section. The image obtained in this step is denoted as the first high-definition image.

[0079] 2) Perform a series of preprocessing operations on the first high-definition image, including: image standardization (such as adjusting to a uniform size and color channel), image enhancement (such as histogram equalization, edge sharpening), denoising processing (such as median filtering, bilateral filtering), etc., to improve the quality and usability of the image. The final first high-definition image is obtained after preprocessing.

[0080] Step S20: obtain a plurality of key points using the scale-invariant feature transformation method according to the first high-definition image, and obtain a microscopic image of each key point to form a first microscopic image set including a plurality of first microscopic images

[0081] The specific implementation is as follows:

[0082] 1) Use the Scale-Invariant Feature Transform (SIFT) algorithm to detect feature points in the first high-definition image, obtaining a plurality of significant key points. The SIFT algorithm can stably detect key points under conditions of image scale, rotation, illumination, etc. These key points often correspond to important structural features in the image.

[0083] 2) For each detected key point, use a high-magnification imaging device (such as an optical microscope, electron microscope, etc.) to capture a local microscopic image of the tumor section tissue at that location. These microscopic images constitute the first microscopic image set.

[0084] Step S30: Obtain the high-definition image of the tumor section at the time of detection, pre-process to obtain the second high-definition image, and according to the plurality of key points, obtain the corresponding microscopic images to form the second microscopic image set, including multiple second microscopic images

[0085] The specific implementation is as follows:

[0086] 1) When it is necessary to detect the biological activity change of the tumor section during the preservation process, use the high-definition camera device again to take a high-definition image of the section. The image obtained in this step is denoted as the second high-definition image.

[0087] 2) Similar to step S20, the SIFT algorithm is used to detect key points in the second high-definition image, and microscopic imaging devices are used to obtain local microscopic images at these key points. These microscopic images constitute the second microscopic image set. It should be noted that the key point positions in the second microscopic image set should be consistent with those in step S20, in order to facilitate subsequent comparative analysis.

[0088] Step S40: Align the first high-definition image and the second high-definition image, and align the first microscopic image and the second microscopic image of each key point

[0089] The specific implementation is as follows:

[0090] 1) Use a feature point-based image registration algorithm, such as the RANSAC algorithm, to perform geometric correction and registration of the first high-definition image and the second high-definition image, eliminating differences in translation, rotation, scaling, etc. during shooting, so that the two high-definition images are aligned in spatial position.

[0091] 2) For each key point, the corresponding first microscopic image and second microscopic image also need to be registered. This can use a local registration algorithm based on image features, such as a local feature matching method based on Scale-Invariant Feature Transform (SIFT). Through this step, it is ensured that the microscopic images of the same key point taken at two time points are completely aligned in spatial position.

[0092] Step S50: Extract the features of the aligned first high-definition image and second high-definition image, denoted as first high-definition features and second high-definition features, respectively; extract the features of the first microscopic image and second microscopic image of each key point, denoted as first microscopic features and second microscopic features, respectively

[0093] The specific implementation is as follows:

[0094] 1) For the first high-definition image and the second high-definition image, the overall features thereof are extracted by using a pre-trained convolutional neural network (such as ResNet50) respectively. These features are denoted as first high-definition features and second high-definition features.

[0095] 2) For the first microscopic image and the second microscopic image corresponding to each key point, the macro-structure features (such as convolutional features based on VGG16), microscopic cell features (such as features based on DenseNet121), and texture features (such as features based on a gray level co-occurrence matrix and a local binary pattern) are extracted respectively. These features are denoted as first microscopic features and second microscopic features.

[0096] Step S60: Input the first high-definition features and second high-definition features, the first microscopic features and second microscopic features of each key point into a pre-trained tumor section preservation process biological activity detection model to obtain a current activity vector of the tumor section and output

[0097] The specific implementation is as follows:

[0098] 1) The tumor section preservation process biological activity detection model adopts a multi-scale feature fusion structure, including a high-definition feature subnetwork, a microscopic macro-structure feature subnetwork, a microscopic microscopic cell feature subnetwork, a microscopic texture feature subnetwork, and a fusion subnetwork.

[0099] 2) The high-definition feature subnetwork is based on a ResNet50 structure and is used to extract the overall features of the high-definition image. The input thereof is the first high-definition features and the second high-definition features, and the output thereof is a high-definition image feature change vector.

[0100] 3) The microscopic macro-structure feature subnetwork is based on a VGG16 structure and is used to extract the macro-structure features of the microscopic image. The input thereof is the first microscopic features and the second microscopic features, and the output thereof is a microscopic macro-structure feature change vector.

[0101] 4) The microscopic microscopic cell feature subnetwork is based on a DenseNet121 structure and is used to extract the microscopic cell features of the microscopic image. The input thereof is the first microscopic features and the second microscopic features, and the output thereof is a microscopic microscopic cell feature change vector.

[0102] 5) The microscopic texture feature subnetwork is based on the feature extraction network of the gray level co-occurrence matrix (GLCM) and the local binary pattern (LBP), and is used for extracting the texture features of the microscopic image, the input of which is the first microscopic feature and the second microscopic feature, and the output is a microscopic texture feature change vector.

[0103] 6) The fusion subnetwork adopts a multi-layer perceptron (MLP) structure, and is used for fusing the features extracted by the subnetworks, the input of which is the high-definition image feature change vector, the microscopic macroscopic feature change vector, the microscopic microscopic feature change vector and the microscopic texture feature change vector, and the output is a current activity vector of the tumor section.

[0104] The second aspect of the present application provides a computer readable storage medium, wherein the computer readable storage medium stores program instructions, and the program instructions are used for executing the tumor section preservation process biological activity detection method.

[0105] The third aspect of the present application provides a tumor section preservation process biological activity detection system, which comprises the computer readable storage medium.

[0106] Specifically, the principle of the present application is as follows:

[0107] Firstly, the method uses a high-definition camera to obtain a high-definition image of the section at the initial time when the section is placed into the preservation device, and uses the SIFT algorithm to detect a plurality of significant key points on the high-definition image. For each key point, a microscopic imaging device is also used to capture a local microscopic image of the section tissue at the position, to form an initial high-definition image set and a microscopic image set. The image data in the initial state records the overall structural features and microscopic cell features of the section.

[0108] When it is necessary to detect the biological activity change of the section in the preservation process, the method again collects a high-definition image and a microscopic image of the section at the time to be detected. By aligning the high-definition images and the microscopic images at the two time points, it can be ensured that the corresponding key points and local areas are completely consistent in spatial position.

[0109] Then, the method extracts the high-definition image features and the microscopic image features in the initial state and the state to be detected, respectively. The high-definition image features include overall structural features, and the microscopic image features include macroscopic structural features, microscopic cell features and texture features, and other multi-scale information. The changes of these features reflect the multi-level structural and functional changes of the section in the preservation process.

[0110] Finally, the extracted high-definition image feature changes and microscopic image feature changes are input into a pre-trained multi-scale feature fusion model. The model learns the mapping relationship between feature changes and biological activity through deep learning. Therefore, the output of the model is the current biological activity level of the slice, providing an important reference for subsequent slice preservation management.

[0111] The above technical effects are mainly due to the innovations of the method in several key aspects: 1) full use of information of two scales of high-definition images and microscopic images to achieve comprehensive characterization of slice structure and function; 2) adoption of precise image registration technology to ensure the time sequence correspondence of features; 3) construction of an analysis model integrating multi-scale feature extraction and deep feature fusion to improve the accuracy of activity evaluation.

[0112] In summary, the tumor slice preservation process biological activity detection method proposed by the present application can fully utilize image analysis and deep learning technology to comprehensively evaluate the multi-scale structure and functional changes of the slice during the preservation process.

[0113] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application.

Claims

1. A method for detecting biological activity in a tumor section preservation process, characterized by, The method comprises the following steps: S10. Obtain a high-definition image of a tumor section just placed in a preservation device, and obtain a first high-definition image after preprocessing; S20. Obtain a plurality of key points using a scale-invariant feature transformation method according to the first high-definition image, and obtain a microscopic image of each key point to form a first microscopic image set comprising a plurality of first microscopic images; S30. Obtain a high-definition image of the tumor section at a detection time, obtain a second high-definition image after preprocessing, and obtain corresponding microscopic images according to the plurality of key points to form a second microscopic image set comprising a plurality of second microscopic images; S40. Align the first high-definition image and the second high-definition image, and align the first microscopic image and the second microscopic image of each key point; S50. Extract features of the aligned first high-definition image and the second high-definition image, denoted as first high-definition features and second high-definition features, respectively; S50. Extract features of the aligned first high-definition image and the second high-definition image, denoted as first high-definition features and second high-definition features, respectively; S60. Input the first high-definition features and the second high-definition features, the first microscopic features and the second microscopic features of each key point into a tumor section preservation process biological activity detection model pre-trained, to obtain a current activity vector of the tumor section and output the current activity vector; The tumor section preservation process biological activity detection model adopts a multi-scale feature fusion structure, comprising a high-definition feature subnetwork, a microscopic macroscopic feature subnetwork, a microscopic microscopic feature subnetwork, a microscopic texture feature subnetwork, and a fusion subnetwork; The high-definition feature subnetwork is used to extract overall features of the high-definition image, the input is the first high-definition features and the second high-definition features, and the output is a high-definition image feature change vector; the structure is a convolutional neural network based on ResNet50; The microscopic macroscopic feature subnetwork is used to extract macroscopic structural features of the microscopic image, the input is the first microscopic features and the second microscopic features, and the output is a microscopic macroscopic feature change vector; the structure is a convolutional neural network based on VGG16; The microscopic microscopic feature subnetwork is used to extract microscopic cell features of the microscopic image, the input is the first microscopic features and the second microscopic features, and the output is a microscopic microscopic feature change vector; the structure is a convolutional neural network based on DenseNet121; The microscopic texture feature subnetwork is used to extract texture features of the microscopic image, the input is the first microscopic features and the second microscopic features, and the output is a microscopic texture feature change vector; the structure is a feature extraction network based on a gray level co-occurrence matrix and a local binary pattern; The fusion subnetwork is used to fuse the features extracted by each subnetwork, the input is the high-definition image feature change vector, the microscopic macroscopic feature change vector, the microscopic microscopic feature change vector, and the microscopic texture feature change vector, and the output is the current activity vector of the tumor section; the structure is a multi-layer perceptron network.

2. The method for detecting bioactivity during tumor slide preservation according to claim 1, characterized in that, The preprocessing includes image standardization, image enhancement, and denoising processing.

3. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores program instructions, and the program instructions are used to execute the tumor section preservation process biological activity detection method of any one of claims 1-2 when running.

4. A system for detecting biological activity in a tumor section preservation process, comprising: The computer-readable storage medium comprises the computer-readable storage medium of claim 3.

Images