Intelligent judgment method and system for lung cancer histopathological typing

Abstract

The invention provides an intelligent judgment method and system for lung cancer tissue pathological typing. The method comprises the following steps: acquiring pathological section image data; preprocessing the pathological section image data to extract feature information in the pathological section image; analyzing the feature information by using a trained deep learning model, wherein the deep learning model is trained based on a large number of labeled lung cancer pathological images; according to an output result of the deep learning model, combining lung cancer pathological typing standards in an expert knowledge base, and intelligently judging a pathological typing category to which the pathological section image belongs; and outputting the pathological typing category and the corresponding confidence degree information to the user. According to the method, the target query statement is generated through the large model, the query intention of a user can be more accurately understood, the corresponding query statement is generated according to the database description information, and therefore the accuracy and correlation of the query result are improved, and the query efficiency is improved by recalling the related sample pairs from the sample set. And the sample pairs are used for performing context learning, so that the system can better understand and process various types of query requests, and can give reasonable response even if the system faces new queries which are not seen before.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of data processing technology, and in particular to an intelligent judgment method and system for lung cancer tissue pathological typing. Background Art

[0002] With the development of information technology, the demand for database query is growing. Especially in the big data environment, how to quickly and accurately generate query statements that conform to the database structure has become a key issue. This not only involves the technical challenge of efficiently retrieving information from massive data, but also requires the system to understand the user's query intent and convert it into specific query statements that the database can execute. Therefore, it is crucial to develop a method that can automatically convert natural language queries into database query statements.

[0003] Currently, most database query generation systems rely on fixed query templates or simple keyword matching. Such systems usually require users to have certain knowledge of database query languages ​​(such as SQL) in order to accurately express query requirements. In addition, there are some machine learning-based methods that try to predict new query statements by learning historical queries, but these methods often require a large amount of labeled data and have limited support for complex queries.

[0004] Traditional query generation technology has the following main problems: first, the fixed template and keyword matching methods are not flexible enough to cope with diverse and complex query requests; second, although machine learning-based methods can improve the accuracy of query generation to a certain extent, due to the lack of understanding of the context, the generated query statements often fail to accurately reflect the user's true intentions; finally, existing solutions usually do not take into account iterative optimization in the query generation process, that is, the system should be able to continuously adjust the generation strategy based on user feedback to improve the relevance and satisfaction of the query results. Summary of the invention

[0005] The embodiments of the present application provide an intelligent judgment method and system for lung cancer tissue pathological typing, which are used to solve the technical problems existing in the above-mentioned background technology.

[0006] In a first aspect, an embodiment of the present application provides methods for obtaining pathological slice image data; preprocessing the pathological slice image data to extract feature information in the pathological slice image; analyzing the feature information using a trained deep learning model, wherein the deep learning model is trained based on a large number of annotated lung cancer pathology images; intelligently determining the pathological classification category to which the pathological slice image belongs based on the output result of the deep learning model and the lung cancer pathological classification standards in an expert knowledge base; and outputting the pathological classification category and corresponding confidence information to a user.

[0007] Further, the preprocessing of the pathological slice image data includes:

[0008] Normalizing the pathological slice image data to meet the requirements of deep learning model input;

[0009] The pathological slice image data is enhanced to increase the diversity of the image and improve the generalization ability of the model.

[0010] Further, the analyzing the feature information using the trained deep learning model includes:

[0011] Use convolutional neural networks (CNNs) to automatically extract low-level and high-level features from pathological images;

[0012] An attention mechanism is used to emphasize important areas in pathological images and reduce the influence of irrelevant information;

[0013] Recurrent neural networks (RNN) or long short-term memory networks (LSTM) are applied to capture the dependencies on spatial locations in pathological images.

[0014] Furthermore, the intelligent determination of the pathological classification category to which the pathological section image belongs by combining the lung cancer pathological classification standard in the expert knowledge base includes:

[0015] Establish a knowledge base containing various lung cancer pathological types and their typical characteristics;

[0016] Comparing the output of the deep learning model with the characteristics of each pathological type in the knowledge base to find the most suitable classification;

[0017] Uncertainty theory is introduced into the judgment process to quantify the degree of uncertainty in model predictions.

[0018] Further, the outputting the pathological classification category and the corresponding confidence information to the user includes:

[0019] Provide a graphical interface to display the pathological classification results and highlight the key areas considered by the model;

[0020] Output detailed reports including pathological classification, confidence score, and recommendations for further examination or treatment.

[0021] Furthermore, it also includes:

[0022] The pathological section image data and the typing results thereof are stored in a database as part of subsequent training data;

[0023] The deep learning model is regularly updated to adapt to new pathological discoveries and technological advances.

[0024] Further, the obtaining of pathological slice image data includes:

[0025] Automatically scan pathological sections to obtain images in digital form;

[0026] Supports manual uploading of existing pathology slide image files.

[0027] Furthermore, before the intelligent determination of the pathological classification category to which the pathological slice image belongs, the method further includes:

[0028] Performing quality control on the pathological slice image data to ensure that the image clarity and integrity meet the analysis requirements;

[0029] An automated workflow is achieved, with no need for human intervention from image acquisition to final pathological classification.

[0030] In a second aspect, an embodiment of the present application provides a computing device, including a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement the method described in the first aspect above.

[0031] In a third aspect, an embodiment of the present application provides a computer storage medium storing a computer program, which, when executed by a computer, implements the method described in the first aspect above.

[0032] In an embodiment of the present invention, pathological slice image data is obtained; the pathological slice image data is preprocessed to extract feature information in the pathological slice image; the feature information is analyzed using a trained deep learning model, the deep learning model is trained based on a large number of annotated lung cancer pathological images; according to the output result of the deep learning model, combined with the lung cancer pathological classification standard in the expert knowledge base, the pathological classification category to which the pathological slice image belongs is intelligently determined; the pathological classification category and the corresponding confidence information are output to the user. By using a large model to generate a target query statement, this method can more accurately understand the user's query intent, and generate corresponding query statements based on the database description information, thereby improving the accuracy and relevance of the query results. By recalling relevant sample pairs from the sample set and using these sample pairs for context learning, the system can better understand and process various types of query requests, and even when faced with new queries that have never been seen before, it can give a reasonable response. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0034] Figure 1 A flowchart of an intelligent judgment method for lung cancer tissue pathological typing provided in an embodiment of the present application;

[0035] Figure 2 A schematic diagram of an intelligent judgment system for lung cancer tissue pathological typing provided in an embodiment of the present application;

[0036] Figure 3 A schematic diagram of a computing device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0037] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0038] Embodiment 1:

[0039] First, the embodiment of the present invention provides an intelligent judgment method for lung cancer tissue pathological typing, which is specifically implemented as follows:

[0040] The specific embodiments are as follows:

[0041] 1. Obtain pathology slide image data: First, obtain a series of digitized pathology slide images from the hospital's pathology laboratory. These images can be high-resolution digital images obtained by scanning glass slides under a microscope.

[0042] 2. Preprocessing of pathological slice image data: Perform necessary preprocessing on the acquired pathological slice image data, which may include grayscale, size standardization, noise removal and other steps to extract key feature information in the pathological slice image. For example, image processing technology can be used to enhance the contrast of cell structures so that the nucleus, cytoplasm and other important structures can be more easily identified.

[0043] 3. Analyze using a trained deep learning model: Next, use a deep learning model that has been trained on a large number of annotated lung cancer pathology images to analyze the preprocessed feature information. This model may be built based on a convolutional neural network (CNN) architecture, which is able to recognize complex patterns in images and has been trained on a large number of lung cancer pathology images with high recognition accuracy. The model training process usually involves annotating a large number of images, marking different types of lung cancer lesions, and then adjusting the network weights through a back-propagation algorithm to optimize the model's classification ability.

[0044] 4. Combine with expert knowledge base for intelligent judgment: After the model analyzes the pathological slice image and generates preliminary results, the system will combine these results with an expert knowledge base containing lung cancer pathological classification standards. This knowledge base contains the experience and guidelines accumulated by pathologists over the years, helping the system to better understand and interpret the output results of the model. For example, the model may identify certain specific cell morphologies or arrangements, and compare them with the standards in the knowledge base, and finally draw conclusions on the pathological classification.

[0045] 5. Output pathological classification and confidence information: Finally, the system will present the analyzed pathological classification and its corresponding confidence information to the user. Confidence information reflects the model's confidence level in the classification result, which can help doctors evaluate the reliability of the diagnosis result. For example, if the model gives a classification of "adenocarcinoma" for a pathological slice image, and the confidence reaches 90%, it means that the model believes that there is a 90% chance that this image belongs to the adenocarcinoma type.

[0046] The entire process is highly automated, which can significantly reduce the workload of pathologists. At the same time, due to the use of deep learning models, its accuracy has also been greatly improved, which helps to improve the accuracy and efficiency of lung cancer pathological diagnosis.

[0047] Embodiment 2:

[0048] The specific embodiments are as follows:

[0049] 1. Obtain pathology slide image data: First, collect digital pathology slide images from clinical trials or pathology laboratories. These images can be high-resolution images generated by digital pathology scanners.

[0050] 2. Preprocessing of pathological slice image data:

[0051] Normalization: To ensure that all image data is in the same scale range, normalize it. For example, you can convert each pixel value into a floating point number between 0 and 1 by dividing it by 255. Such normalization helps speed up model training and ensures data consistency.

[0052] Enhancement processing: In order to increase the generalization ability of the model, the images are enhanced, such as rotation, scaling, flipping, brightness adjustment, etc., to simulate different shooting angles and conditions, thereby generating more training samples to cover more situations.

[0053] 3. Use the trained deep learning model for analysis:

[0054] Use convolutional neural network (CNN): CNN can automatically extract low-level features such as edges, textures, shapes, and other high-level features in pathological images, as well as the size, morphology, and distribution of cell nuclei. Through multi-layer convolution, pooling, and full connection operations, CNN can learn multi-level feature representations from local to global levels.

[0055] Adopting attention mechanism: Adding attention mechanism module on the basis of CNN enables the model to focus on key areas in pathological images, such as where abnormal cells gather, and ignore background noise or irrelevant parts. This helps to improve the diagnostic accuracy of the model.

[0056] Applying recurrent neural networks (RNN) or long short-term memory networks (LSTM): When analyzing the spatial dependencies of pathological images, RNN or LSTM can capture the relative position information between cells, which is very important for understanding the structure of tumor tissue. Through memory units, these models are able to retain long-term dependencies, thereby better understanding the spatial layout in the image.

[0057] 4. Combine expert knowledge base for intelligent judgment:

[0058] Establish a knowledge base: Create a knowledge base containing a variety of lung cancer pathological types and their typical characteristics, such as squamous cell carcinoma, adenocarcinoma, small cell lung cancer, etc. Each type is accompanied by a detailed description, including common cell morphology, arrangement, staining characteristics, etc.

[0059] Compare model output with knowledge base features: The output of the deep learning model (such as cell features, tissue structure, etc.) is compared with the pathology type features in the knowledge base to find the best matching classification. For example, if the model detects glandular structures and mucus secretions, it may match the characteristics of adenocarcinoma.

[0060] Introducing uncertainty theory: In the judgment process, uncertainty estimation is introduced through Bayesian neural networks or other probabilistic models to evaluate the credibility of the model prediction results. This can not only help doctors understand the reliability of model predictions, but also remind doctors that further inspection or manual review is required when the model is uncertain.

[0061] Through the above steps, the present invention provides an intelligent judgment method, which can not only efficiently analyze pathological slice images, but also make more accurate pathological classification judgments in combination with expert knowledge, greatly improving the accuracy and efficiency of lung cancer pathological diagnosis.

[0062] Embodiment three:

[0063] The specific embodiments are as follows:

[0064] 1. Obtain pathological slice image data:

[0065] Automatic scanning of pathology slides: Use digital pathology scanners to automatically scan pathology slides and generate high-resolution digital images. This process ensures the quality of the image while reducing errors caused by human operation.

[0066] Manually upload image files: For existing pathology slide images, the system allows users to manually upload image files through a graphical interface to facilitate the integration of existing data.

[0067] 2. Image quality control:

[0068] Before the image enters the subsequent processing flow, quality control is performed to check whether the image is clear, complete, and free of blur, missing or other problems. At this stage, simple image processing techniques such as edge detection can be used to verify whether the image is suitable for analysis. Only images that meet certain standards will be sent to the next step.

[0069] 3. Automated Workflow:

[0070] From image acquisition to pathological classification, the entire process is fully automated without human intervention. This means that the digitization of pathological sections, preprocessing, feature extraction, model analysis, and final result output are all completed automatically by the system without the need for additional manual operations.

[0071] 4. Intelligent judgment of pathological classification categories:

[0072] After preprocessing and feature extraction, a pre-trained deep learning model is used to analyze the pathological slice image, determine which pathological type it belongs to, and give a corresponding confidence score. This process may involve the application of multiple technologies such as convolutional neural network (CNN), attention mechanism, recurrent neural network (RNN) or long short-term memory network (LSTM).

[0073] 5. Output the results to the user:

[0074] Graphical interface display: The pathological classification results are displayed through a user-friendly graphical interface. The areas that the model considers important will be highlighted to facilitate users to quickly locate and understand.

[0075] Detailed report output: The system will generate a detailed electronic report that includes pathological classification information and confidence scores, as well as recommendations for further examination or treatment based on the model's output. This report can be sent directly to the user or saved in the system for the user to download and view.

[0076] 6. Data storage and model update:

[0077] Data storage: The processed pathological slide image data and its classification results are stored in the database, which can be used as part of the new data set for future model training.

[0078] Model update: The model is regularly updated based on the latest pathology research results and technological advances to maintain its advanced performance and accuracy.

[0079] Through the above steps, this method not only provides a one-stop solution from pathological slice image acquisition to intelligent analysis, but also ensures the continuous improvement of the system so that it can be continuously optimized with the development of medical research.

[0080] Embodiment 4:

[0081] The present application also provides a computer storage medium storing a computer program, wherein the computer program can achieve the above-mentioned Figure 1 The method of the illustrated embodiment.

[0082] Embodiment five:

[0083] The embodiment of the present application also provides an intelligent judgment system for lung cancer tissue pathology typing based on cloud computing, characterized in that it includes the method and apparatus described in Embodiment 1 and Embodiment 2, and is connected to a remote server cluster via a network to achieve distributed storage and processing of lower respiratory tract infection data.

[0084] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0085] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0086] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0087] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. 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 application.

Claims

1. An intelligent judgment method for lung cancer tissue pathological typing, characterized in that: include: Acquire pathological slice image data; Preprocessing the pathological slice image data to extract characteristic information in the pathological slice image; Analyzing the feature information using a trained deep learning model, where the deep learning model is trained based on a large number of annotated lung cancer pathology images; According to the output results of the deep learning model, combined with the lung cancer pathology classification standards in the expert knowledge base, intelligently determine the pathology classification category to which the pathology section image belongs; The pathological classification category and the corresponding confidence information are output to the user.

2. The method according to claim 1, characterized in that The preprocessing of the pathological slice image data comprises: Normalizing the pathological slice image data to meet the requirements of deep learning model input; The pathological slice image data is enhanced to increase the diversity of the image and improve the generalization ability of the model.

3. The method according to claim 1, characterized in that The analyzing the feature information by using the trained deep learning model includes: Use convolutional neural networks (CNNs) to automatically extract low-level and high-level features from pathological images; An attention mechanism is used to emphasize important areas in pathological images and reduce the influence of irrelevant information; Recurrent neural networks (RNN) or long short-term memory networks (LSTM) are applied to capture the dependencies on spatial locations in pathological images.

4. The method according to claim 1, characterized in that The intelligent determination of the pathological classification category to which the pathological section image belongs in combination with the lung cancer pathological classification standard in the expert knowledge base includes: Establish a knowledge base containing various lung cancer pathological types and their typical characteristics; Comparing the output of the deep learning model with the characteristics of each pathological type in the knowledge base to find the most suitable classification; Uncertainty theory is introduced into the judgment process to quantify the degree of uncertainty in model predictions.

5. The method according to claim 1, characterized in that The step of outputting the pathological classification category and the corresponding confidence information to the user comprises: Provide a graphical interface to display the pathological classification results and highlight the key areas considered by the model; Output detailed reports including pathological classification, confidence score, and recommendations for further examination or treatment.

6. The method according to claim 1, characterized in that Also includes: The pathological section image data and the typing results thereof are stored in a database as part of subsequent training data; The deep learning models are regularly updated to adapt to new pathology discoveries and technological advances.

7. The method according to claim 1, characterized in that The obtaining of pathological slice image data comprises: Automatically scan pathological sections to obtain images in digital form; Supports manual uploading of existing pathology slide image files.

8. The method according to claim 1, characterized in that Before the intelligent determination of the pathological classification category to which the pathological slice image belongs, the method further includes: Performing quality control on the pathological slice image data to ensure that the image clarity and integrity meet the analysis requirements; An automated workflow is achieved, with no need for human intervention from image acquisition to final pathological classification.

9. An electronic device, comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to run the computer program to perform the method according to any one of claims 1 to 8.

10. A computer storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images