CN112884724B - Intelligent judgment method and system for lung cancer histopathological typing - Google Patents
Intelligent judgment method and system for lung cancer histopathological typing Download PDFInfo
- Publication number
- CN112884724B CN112884724B CN202110144607.7A CN202110144607A CN112884724B CN 112884724 B CN112884724 B CN 112884724B CN 202110144607 A CN202110144607 A CN 202110144607A CN 112884724 B CN112884724 B CN 112884724B
- Authority
- CN
- China
- Prior art keywords
- lung cancer
- model
- slice
- classification
- histopathological
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30061—Lung
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30096—Tumor; Lesion
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Investigating Or Analysing Biological Materials (AREA)
Abstract
The invention discloses an intelligent judgment method and system for lung cancer histopathological typing, wherein the method comprises the following steps: acquiring lung cancer tissue slice data meeting preset requirements, and then preprocessing to obtain classified lung cancer samples; respectively constructing a slice tile model and a slice aggregation model according to the lung cancer sample; after the model is built, model training, verification and testing are respectively carried out to form a lung cancer quantitative histology classification evaluation system; and inputting the lung cancer histopathological section to be classified to the lung cancer quantitative histology classification evaluation system to obtain an artificial intelligent diagnosis report opinion book and a corresponding lung cancer histopathological classification result. According to the invention, the lung cancer quantitative histology classification evaluation system is constructed to carry out intelligent auxiliary pathological typing, so that the repeated labor of a pathological diagnosis doctor is reduced, the efficiency of pathological typing diagnosis is effectively improved on the premise of ensuring the accuracy of pathological diagnosis, and the digitization and the intellectualization of the pathological subject development are promoted.
Description
Technical Field
The invention relates to the technical field of computer-aided diagnosis, in particular to an intelligent judgment method and system for lung cancer histopathological typing.
Background
Pathological diagnosis is the fundamental link of individual treatment of patients and is the 'gold standard' for judging cancers. For lung cancer, the multidisciplinary classification system standard of research results in fields such as oncology, molecular biology, pathology, radiology, surgery and the like is integrated, so that pathological classification can be better served for clinical practice and clinical/basic research. According to the classification of WHO lung tumors (2015) and the comprehensive and detailed histological diagnosis mode advocated by the classification, a pathologist classifies the non-small cell cancers into squamous cell lung cancer, adenocarcinoma lung cancer and large cell lung cancer on the basis of the classification of the existing small cell cancers and the non-small cell lung cancers. Meanwhile, a plurality of morphological characteristics are proved to have important prognostic significance on lung adenocarcinoma.
The prior art classifies invasive lung adenocarcinoma mainly according to the main growth mode of tumors, and semi-quantitatively evaluates the proportion of four different growth modes, namely, mural-like, acinar-like, papillary and solid, in 5% increment, and the diagnosis is described when the proportion reaches 5%. The pathologist finally gives a classification diagnosis recommendation by repeatedly reading the lung tissue biopsy under the microscope (5-10 sections per patient per day).
However, in the research and practice process of the prior art, the inventor of the present invention finds that the current lung cancer pathological tissue diagnosis is mainly a manual diagnosis mode, the diagnosis automation degree is low, and the diagnosis time is long. And because the number index of pathological doctors in every ten thousand people in China is about 0.12, which is far lower than 0.52 and 0.55 in the United states and Europe, the index can not meet the requirement of lung cancer diagnosis increase in clinical work, so that the burden and responsibility of pathological doctors are heavy, and the misdiagnosis rate and the missed diagnosis rate are easy to increase.
Disclosure of Invention
The technical problem to be solved by the embodiments of the present invention is to provide an intelligent determination method and system for pathological typing of lung cancer tissue, which can effectively improve the efficiency of pathological typing diagnosis on the premise of ensuring the accuracy of pathological diagnosis.
In order to solve the above problem, a first aspect of the embodiments of the present application provides an intelligent determination method for lung cancer histopathological typing, which at least includes the following steps:
acquiring lung cancer tissue slice data meeting preset requirements, and then preprocessing to obtain classified lung cancer samples;
respectively constructing a slice tile model and a slice aggregation model according to the lung cancer sample;
after the model is built, model training, verification and testing are respectively carried out to form a lung cancer quantitative histology classification evaluation system;
and inputting the lung cancer histopathological section to be classified to the lung cancer quantitative histology classification evaluation system to obtain an artificial intelligent diagnosis report opinion book and a corresponding lung cancer histopathological classification result.
In a possible implementation manner of the first aspect, the method for intelligently determining a pathological typing of a lung cancer further includes:
and randomly dividing the lung cancer samples according to a preset proportion to obtain a training set, a verification set and a test set corresponding to the tile model.
In a possible implementation manner of the first aspect, the constructing of the slicing tile model specifically includes:
and (3) building a plurality of convolutional neural network models by combining a basic neural network model and a pathological image recognition technology and optimizing the layer number, the depth and the super parameters of the basic neural network model.
In a possible implementation manner of the first aspect, the training, verifying, and testing of the slice tile model specifically includes:
respectively inputting a training set, a verification set and a test set corresponding to the tile model to a plurality of built convolutional neural network models for training, verifying and testing;
and comparing the prediction accuracy of the plurality of convolutional neural network models based on the test set, and selecting the convolutional neural network model with the highest prediction accuracy as the optimal classification model.
In a possible implementation manner of the first aspect, the constructing of the slice aggregation model specifically includes:
and (3) extracting the characteristics of the slice tile model to form a slice heat map, quantifying according to whether each region of the slice heat map has a count, and performing standard definition on a quantification result by using the label of a pathologist.
In a possible implementation manner of the first aspect, the model training, verifying, and testing of the slice aggregation model specifically includes:
extracting the characteristics of the tile model to form a tile heat map, marking the tile heat map, and performing model training as a training set of the tile aggregation model;
and taking the marked slice WSI image as a verification set and a test set of the slice aggregation model.
In a possible implementation manner of the first aspect, the preprocessing specifically includes:
performing two-classification labeling on each cell in the WSI image by drawing the cells of the WSI image in the lung cancer tissue slice data;
and performing three-classification evaluation on the lung cancer database subjected to the two-classification evaluation, performing two-classification evaluation on each cell in the lung adenocarcinoma, and performing four-classification evaluation on the lung adenocarcinoma database.
In a possible implementation manner of the first aspect, the method for intelligently judging the pathological type of the lung cancer further includes:
and carrying out external evaluation on the artificial intelligent diagnosis report opinion book, and determining evaluation index data of the lung cancer quantitative histology classification evaluation system according to an external evaluation result.
In a possible implementation manner of the first aspect, the method for intelligently judging the pathological type of the lung cancer further includes:
and acquiring the external evaluation result in real time, and correcting and updating the lung cancer quantitative histology classification evaluation system according to the external evaluation result.
A second aspect of the embodiments of the present application further provides an intelligent determination system for lung cancer histopathological typing, including:
the pretreatment module is used for carrying out pretreatment after lung cancer tissue slice data meeting the preset requirements are collected, so as to obtain a lung cancer sample after classification and marking;
the model construction module is used for respectively constructing a slice tile model and a slice aggregation model according to the lung cancer sample;
the evaluation system construction module is used for respectively carrying out model training, verification and testing after the model is constructed to form a lung cancer quantitative histology classification evaluation system;
and the classification evaluation module is used for inputting the lung cancer histopathological section to be classified to the lung cancer quantitative histology classification evaluation system to obtain an artificial intelligent diagnosis report opinion book and a corresponding lung cancer histopathological classification result.
The embodiment of the invention has the following beneficial effects:
the embodiment of the invention provides an intelligent judgment method and system for lung cancer histopathological typing, wherein the method comprises the following steps: acquiring lung cancer tissue slice data meeting preset requirements, and then preprocessing to obtain classified lung cancer samples; respectively constructing a slice tile model and a slice aggregation model according to the lung cancer sample; after the model is built, model training, verification and testing are respectively carried out to form a lung cancer quantitative histology classification evaluation system; and inputting the lung cancer histopathological section to be classified to the lung cancer quantitative histology classification evaluation system to obtain an artificial intelligent diagnosis report opinion book and a corresponding lung cancer histopathological classification result.
Compared with the prior art, the embodiment of the invention can carry out intelligent auxiliary pathological typing by constructing a lung cancer quantitative histology classification evaluation system, reduce the repeated labor of pathological diagnosis doctors, effectively improve the efficiency of pathological typing diagnosis and promote the digitization and the intellectualization of pathological subject development on the premise of ensuring the accuracy of pathological diagnosis.
Drawings
Fig. 1 is a schematic flowchart of an intelligent determination method for pathological typing of lung cancer according to a first embodiment of the present invention;
FIG. 2 is a schematic flowchart of a lung cancer quantitative histological classification evaluation system according to a first embodiment of the present invention;
fig. 3 is a schematic flowchart of another intelligent determination method for pathological typing of lung cancer according to the first embodiment of the present invention;
FIG. 4 is a schematic structural diagram of an intelligent determination system for histopathological typing of lung cancer according to a second embodiment of the present invention;
fig. 5 is a schematic structural diagram of another intelligent determination system for histopathological typing of lung cancer according to a second embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
In the description of the present application, it is to be understood that the terms "first", "second", and the like are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implying any number of technical features indicated. Thus, a feature defined as "first," "second," etc. may explicitly or implicitly include one or more of that feature. In the description of the present application, "a plurality" means two or more unless otherwise specified.
Firstly, the application scenarios provided by the invention are introduced, such as constructing a lung cancer quantitative histology classification evaluation system for intelligent auxiliary pathological typing.
The first embodiment of the present invention:
please refer to fig. 1-3.
As shown in fig. 1, the present embodiment provides an intelligent determination method for pathological typing of lung cancer, which at least includes the following steps:
and S1, collecting lung cancer tissue slice data meeting the preset requirements, and then preprocessing the lung cancer tissue slice data to obtain a lung cancer sample after classification and marking.
In a preferred embodiment, the pretreatment specifically includes:
performing two-classification labeling on each cell in the WSI image by drawing the cells of the WSI image in the lung cancer tissue slice data;
and performing three-classification evaluation on the lung cancer database subjected to the two-classification evaluation, performing two-classification evaluation on each cell in the lung adenocarcinoma, and performing four-classification evaluation on the lung adenocarcinoma database.
Specifically, in step S1, the present embodiment collects lung cancer tissue section data that meets the requirements of the pathology department of hospitals (i.e., that can be judged by histological classification). The pathologist performs two classification labels on the WSI image frames, each tile (small grid) of the WSI image, for example, normal VS specific disease species, namely lung adenocarcinoma/lung squamous carcinoma, and performs three classification evaluations on a lung cancer database according to WHO lung cancer histological classification standards (2015), including lung adenocarcinoma, lung squamous carcinoma and normal. And performing a dichotomous labeling of each tile of lung adenocarcinoma according to WHO lung cancer histological classification criteria (2015), such as normal VS specific disease species, i.e. parietal/alveolar/papillary/substantial, and performing a quartile assessment of the lung adenocarcinoma database according to WHO lung cancer histological classification criteria (2015), including parietal, alveolar, papillary, and substantial.
And S2, respectively constructing a slice tile model and a slice aggregation model according to the lung cancer sample.
In a preferred embodiment, the constructing of the slice tile model specifically includes:
and (3) building a plurality of convolutional neural network models by combining a basic neural network model and a pathological image recognition technology and optimizing the layer number, the depth and the super parameters of the basic neural network model.
In a preferred embodiment, the construction of the slice aggregation model specifically includes:
and (3) extracting the characteristics of the slice tile model to form a slice heat map, quantifying according to whether each region of the slice heat map has a count, and performing standard definition on a quantification result by using the label of a pathologist.
Specifically, for step S2, for the construction of the tile model, based on the existing basic network model, in combination with the characteristics of pathological image recognition, the number of layers, depth and hyper-parameter tuning are fully considered, and a plurality of convolutional neural network models (including se _ rest 101, se _ rest 152, initiation _ v4, senet154, etc.) are constructed; for the construction of the slice aggregation model, the characteristics of the slice tile model are extracted to form a whole slice heat map, and quantification is carried out according to whether the counting appears in each area or not, and the standard definition is carried out on the quantification result of the invention by labeling of a pathologist.
And S3, respectively training, verifying and testing the models after the models are constructed to form a lung cancer quantitative histology classification evaluation system.
In a preferred embodiment, the training, verifying and testing of the slicing tile model specifically includes:
respectively inputting a training set, a verification set and a test set corresponding to the tile model to a plurality of built convolutional neural network models for training, verifying and testing;
and comparing the prediction accuracy of the plurality of convolutional neural network models based on the test set, and selecting the convolutional neural network model with the highest prediction accuracy as the optimal classification model.
In a preferred embodiment, the model training, verifying and testing of the slice aggregation model specifically includes:
extracting the characteristics of the tile model to form a tile heat map, marking the tile heat map, and performing model training as a training set of the tile aggregation model;
and taking the marked slice WSI image as a verification set and a test set of the slice aggregation model.
Specifically, for step S3, after building a plurality of convolutional neural network models, inputting the data set to perform training, verification, and testing, and comparing the prediction accuracy of each model test set to obtain a classification model with the optimal effect; after the verification and test of the verification set data, the slice aggregation model finally forms a DM-pathology lung cancer quantitative histological classification evaluation system.
As shown in fig. 2, the DM-pathology lung cancer quantitative histological classification evaluation system procedure includes two panels: slice tile model and slice aggregation model. The categories used primarily include: normal/squamous lung carcinoma/adenocarcinoma of the lung (parietal, alveolar, papillary, parenchymal).
And S4, inputting the lung cancer histopathological section to be classified to the lung cancer quantitative histology classification evaluation system to obtain an artificial intelligent diagnosis report opinion book and a corresponding lung cancer histopathological classification result.
Specifically, in step S4, after the lung cancer histopathological section to be pathologically diagnosed and typed is inputted to the above-mentioned DM-pathology lung cancer quantitative histological classification evaluation system, the artificial intelligence diagnosis report opinion and the corresponding lung cancer histopathological classification result are obtained after automatic diagnosis, and the time is not more than 1 min/piece.
In a preferred embodiment, the method for intelligently judging the pathological type of lung cancer further includes:
and randomly dividing the lung cancer samples according to a preset proportion to obtain a training set, a verification set and a test set corresponding to the tile model.
Specifically, all lung cancer samples are divided into a training set, a validation set and a testing set according to a ratio of 7:2: 1. For the slice tile model, randomly and proportionally dividing the labeled tile images of the lung cancer sample into a training set for parameter training, a verification set for verifying the model and a test set for predicting accuracy.
For the slice aggregation model, the features of the slice tile model are extracted to form a heat map and labeled as a training set, and the labeled other slice WSI images are used as a verification set and a test set.
Wherein the total sample size ratio of the tile model to the aggregation model is 500: 1.
in a preferred embodiment, as shown in fig. 3, the method for intelligently determining the pathological type of lung cancer further includes:
s5, carrying out external evaluation on the artificial intelligent diagnosis report opinion book, and determining evaluation index data of the lung cancer quantitative histology classification evaluation system according to an external evaluation result.
And S6, acquiring the external evaluation result in real time, and correcting and updating the lung cancer quantitative histology classification evaluation system according to the external evaluation result.
Specifically, in steps S5 and S6, the lung cancer quantitative histology classification evaluation system preliminarily compares the evaluation of pathologists (external test) after acquiring lung cancer histopathological sections and viewing the opinion book of artificial intelligent diagnosis report, and determines the evaluation index of the evaluation system: accuracy, sensitivity, specificity. And subsequently, the lung cancer quantitative histology classification evaluation system can be corrected and updated by collecting the evaluation results of a pathologist in real time and combining each evaluation index, so that the accuracy, sensitivity and specificity of the classification evaluation system are improved.
The intelligent judgment method for pathological typing of lung cancer provided by the embodiment comprises the following steps: acquiring lung cancer tissue slice data meeting preset requirements, and then preprocessing to obtain classified lung cancer samples; respectively constructing a slice tile model and a slice aggregation model according to the lung cancer sample; after the model is built, model training, verification and testing are respectively carried out to form a lung cancer quantitative histology classification evaluation system; and inputting the lung cancer histopathological section to be classified to the lung cancer quantitative histology classification evaluation system to obtain an artificial intelligent diagnosis report opinion book and a corresponding lung cancer histopathological classification result.
Compared with the prior art, the intelligent judgment method for lung cancer histopathological typing provided by the embodiment of the invention trains, verifies and tests the data labeled by prophase pathological doctors through a deep learning framework, realizes classification and grading of lung cancer tissue slice WSI images, refines qualitative or quantitative evaluation of various histological features of lung cancer, and more accurately classifies histological prognosis. The system can promote the working efficiency and the comprehensive level of pathological diagnosis, solves the current situation of great difference of the current medical level of China due to the network sharing of intelligent digital pathology, promotes the reasonable distribution of medical resources and implements a graded diagnosis and treatment system, helps doctors to make quick and accurate pathological diagnosis decisions, and simultaneously improves the pathological diagnosis level of the medical underdeveloped areas of China, thereby promoting the development digitization and the intellectualization of pathological subjects.
Second embodiment of the invention:
please refer to fig. 4 and 5.
As shown in fig. 4, the present embodiment provides an intelligent determination system for histopathological typing of lung cancer, including:
and the preprocessing module 100 is configured to acquire lung cancer tissue slice data meeting preset requirements and then preprocess the acquired lung cancer tissue slice data to obtain a classified lung cancer sample.
Specifically, for the preprocessing module 100, lung cancer tissue slice data that meets the requirements of the pathology department of hospitals (i.e., that can be judged by histological classification) is collected. The pathologist performs two classification labels on the WSI image frames, each tile (small grid) of the WSI image, for example, normal VS specific disease species, namely lung adenocarcinoma/lung squamous carcinoma, and performs three classification evaluations on a lung cancer database according to WHO lung cancer histological classification standards (2015), including lung adenocarcinoma, lung squamous carcinoma and normal. And the lung adenocarcinoma is given two classification labels per tile according to WHO lung cancer histological classification criteria (2015), such as normal VS specific disease species, i.e. parietal/acinar/papillary/solidity, and four classification assessments are made on the lung adenocarcinoma database according to WHO lung cancer histological classification criteria (2015) including parietal, acinar, papillary and solidity.
And the model building module 200 is used for respectively building a slice tile model and a slice aggregation model according to the lung cancer sample.
Specifically, for the model construction module 200, for the construction of the tile model, based on the existing basic network model, in combination with the characteristics of pathological image recognition, the number of layers, depth and super-parameter tuning are fully considered, and a plurality of convolutional neural network models (including se _ rest 101, se _ rest 152, initiation _ v4, sensing 154, and the like) are constructed; for the construction of the slice aggregation model, the characteristics of the slice tile model are extracted to form a whole slice heat map, and quantification is carried out according to whether the counting appears in each area or not, and the standard definition is carried out on the quantification result of the invention by labeling of a pathologist.
And the evaluation system building module 300 is used for respectively performing model training, verification and testing after the model building is completed, so as to form a lung cancer quantitative histology classification evaluation system.
Specifically, for the evaluation system building module 300, after building a plurality of convolutional neural network models, the data sets are input for training, verification and testing, and the prediction accuracy of each model test set is compared to obtain a classification model with the optimal effect; after the verification set data verification and test of the slice aggregation model, a DM-pathway lung cancer quantitative histological classification evaluation system is finally formed.
And the classification evaluation module 400 is used for inputting the lung cancer histopathological section to be classified to the lung cancer quantitative histology classification evaluation system to obtain an artificial intelligent diagnosis report opinion book and a corresponding lung cancer histopathological classification result.
Specifically, for the classification evaluation module 400, after the lung cancer histopathological section to be pathologically diagnosed and classified is inputted to the above-mentioned DM-pathology lung cancer quantitative histological classification evaluation system, the artificial intelligence diagnosis report opinion and the corresponding lung cancer histopathological classification result are obtained after automatic diagnosis, and the time is not more than 1 min/piece.
In a preferred embodiment, as shown in fig. 5, the intelligent determination system for histopathological typing of lung cancer further includes:
and the external testing module 500 is used for performing external evaluation on the artificial intelligent diagnosis report opinion book and determining evaluation index data of the lung cancer quantitative histology classification evaluation system according to an external evaluation result.
And a correction updating module 600, configured to collect the external evaluation result in real time, and correct and update the lung cancer quantitative histology classification evaluation system according to the external evaluation result.
Specifically, for the external test module 500 and the correction update module 600, the lung cancer quantitative histology classification evaluation system preliminarily compares the evaluation (external test) of a pathologist after acquiring lung cancer histopathological sections and viewing the opinion book of the artificial intelligent diagnosis report, and determines the evaluation index of the evaluation system: accuracy, sensitivity, specificity. And subsequently, the lung cancer quantitative histology classification evaluation system can be corrected and updated by collecting the evaluation results of a pathologist in real time and combining each evaluation index, so that the accuracy, sensitivity and specificity of the classification evaluation system are improved.
The present embodiment provides an intelligent judgment system for lung cancer histopathological typing, including: the pretreatment module is used for carrying out pretreatment after lung cancer tissue slice data meeting the preset requirements are collected, so as to obtain a lung cancer sample after classification and marking; the model construction module is used for respectively constructing a slice tile model and a slice aggregation model according to the lung cancer sample; the evaluation system construction module is used for respectively carrying out model training, verification and testing after the model is constructed to form a lung cancer quantitative histology classification evaluation system; and the classification evaluation module is used for inputting the lung cancer histopathological section to be classified to the lung cancer quantitative histology classification evaluation system to obtain an artificial intelligent diagnosis report opinion book and a corresponding lung cancer histopathological classification result.
This embodiment can carry out intelligent supplementary pathological typing through constructing lung cancer quantitative histology classification evaluation system, reduces pathological diagnosis doctor's repeated work, under the prerequisite of guaranteeing pathological diagnosis's rate of accuracy, effectively improves the diagnostic efficiency of pathological typing, promotes the digitization and the intellectuality of pathological subject development.
In the above embodiments of the present invention, the description of each embodiment has its own emphasis, and reference may be made to the related description of other embodiments for parts that are not described in detail in a certain embodiment.
In the embodiments provided in the present application, it should be understood that the disclosed technology can be implemented in other ways. The above-described system embodiments are merely illustrative, and for example, the division of the modules may be a logical division, and in actual implementation, there may be another division, for example, multiple modules or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, units or modules, and may be in an electrical or other form.
The modules described as separate parts may or may not be physically separate, and parts displayed as modules may or may not be physical modules, may be located in one place, or may be distributed on a plurality of modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, functional modules in the embodiments of the present invention may be integrated into one processing module, or each of the modules may exist alone physically, or two or more modules are integrated into one module. The integrated module can be realized in a hardware mode, and can also be realized in a software functional module mode.
The foregoing is directed to the preferred embodiment of the present invention, and it is understood that various changes and modifications may be made by one skilled in the art without departing from the spirit of the invention, and it is intended that such changes and modifications be considered as within the scope of the invention.
It will be understood by those skilled in the art that all or part of the processes in the methods of the above embodiments may be implemented by instructing relevant hardware through a computer program, and the program may be installed in a computer, and after the program is installed, the processes including the embodiments of the methods may be executed.
Claims (4)
1. An intelligent judgment method for lung cancer histopathological typing is characterized by at least comprising the following steps:
acquiring lung cancer tissue slice data meeting preset requirements, and then preprocessing to obtain classified lung cancer samples; respectively constructing a slice tile model and a slice aggregation model according to the lung cancer sample;
after the model is built, model training, verification and testing are respectively carried out to form a lung cancer quantitative histology classification evaluation system;
inputting the lung cancer histopathological section to be classified to the lung cancer quantitative histology classification evaluation system to obtain artificial intelligence
The diagnostic report opinion book and the corresponding lung cancer histopathology typing result;
the intelligent judgment method further comprises the following steps: randomly dividing the lung cancer sample according to a preset proportion to obtain a training set, a verification set and a test set corresponding to the tile model;
the pretreatment specifically comprises the following steps: performing two-classification labeling on each cell in the WSI image by drawing the cells in the lung cancer tissue section data, wherein the two classifications comprise normal and specific disease species, and the specific disease species comprise lung adenocarcinoma or lung squamous carcinoma; performing three-classification evaluation on the lung cancer database labeled by the two classifications, wherein the three classifications comprise lung adenocarcinoma, lung squamous carcinoma and normal; performing two classification labels on each cell in the lung adenocarcinoma, wherein the two classification labels comprise normal and specific disease species; performing four-classification evaluation on the lung adenocarcinoma database, wherein the four classifications comprise parietal, acinar, papillary and substantivity;
the construction of the slice tile model specifically comprises the following steps: building a plurality of convolutional neural network models by combining a basic neural network model and a pathological image identification technology and optimizing the layer number, the depth and the super parameters of the basic neural network model;
the model training, verification and test of the slice tile model specifically comprise: respectively inputting a training set, a verification set and a test set corresponding to the tile model to a plurality of built convolutional neural network models for training, verifying and testing;
comparing the prediction accuracy rates of the plurality of convolutional neural network models based on the test set, and selecting the convolutional neural network model with the highest prediction accuracy rate as an optimal classification model;
the construction of the slice aggregation model specifically comprises the following steps: extracting the characteristics of the slice tile model to form a slice heat map, quantifying according to whether each region of the slice heat map has a count, and performing standard definition on a quantitative result according to the label of a pathologist;
the model training, verifying and testing of the slice aggregation model specifically comprises the following steps: extracting the characteristics of the slice tile model to form a slice heat map, marking the slice heat map, and performing model training as a training set of the slice aggregation model; and taking the marked slice WSI image as a verification set and a test set of the slice aggregation model.
2. The intelligent judgment method for lung cancer histopathological typing according to claim 1, further comprising: and carrying out external evaluation on the artificial intelligent diagnosis report opinion book, and determining evaluation index data of the lung cancer quantitative histology classification evaluation system according to an external evaluation result.
3. The intelligent judgment method for lung cancer histopathological typing according to claim 2, further comprising: and acquiring the external evaluation result in real time, and correcting and updating the lung cancer quantitative histology classification evaluation system according to the external evaluation result.
4. An intelligent judgment system for histopathological typing of lung cancer by the intelligent judgment method according to any one of claims 1 to 3, comprising: the pretreatment module is used for carrying out pretreatment after lung cancer tissue slice data meeting the preset requirements are collected, so as to obtain a lung cancer sample after classification and marking; the model building module is used for respectively building a slice tile model and a slice aggregation model according to the lung cancer sample; the evaluation system construction module is used for respectively carrying out model training, verification and testing after the model is constructed to form a lung cancer quantitative histology classification evaluation system;
and the classification evaluation module is used for inputting the lung cancer histopathological section to be classified to the lung cancer quantitative histology classification evaluation system to obtain an artificial intelligent diagnosis report opinion book and a corresponding lung cancer histopathological classification result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110144607.7A CN112884724B (en) | 2021-02-02 | 2021-02-02 | Intelligent judgment method and system for lung cancer histopathological typing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110144607.7A CN112884724B (en) | 2021-02-02 | 2021-02-02 | Intelligent judgment method and system for lung cancer histopathological typing |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112884724A CN112884724A (en) | 2021-06-01 |
| CN112884724B true CN112884724B (en) | 2022-06-03 |
Family
ID=76055819
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110144607.7A Active CN112884724B (en) | 2021-02-02 | 2021-02-02 | Intelligent judgment method and system for lung cancer histopathological typing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112884724B (en) |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114387214B (en) * | 2021-12-09 | 2024-09-06 | 中山大学附属第一医院 | Artificial intelligent lung cancer pathological immunophenotype prediction and auxiliary parting system |
| CN115082718A (en) * | 2022-05-06 | 2022-09-20 | 清华大学 | Glioma grading method, device, equipment and medium based on histopathology image |
| CN115132375A (en) * | 2022-06-17 | 2022-09-30 | 广州智睿医疗科技有限公司 | Thyroid disease pathological analysis module |
| CN116309368B (en) * | 2023-02-21 | 2023-11-14 | 北京透彻未来科技有限公司 | Lung cancer pathological diagnosis system based on deep migration learning |
| CN117152509B (en) * | 2023-08-28 | 2024-04-30 | 北京透彻未来科技有限公司 | Stomach pathological diagnosis and typing system based on cascade deep learning |
| CN117173485B (en) * | 2023-09-18 | 2024-02-13 | 西安交通大学医学院第二附属医院 | Intelligent classification system method and system for lung cancer tissue pathological images |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109754879A (en) * | 2019-01-04 | 2019-05-14 | 湖南兰茜生物科技有限公司 | A kind of lung cancer computer aided detection method and system based on deep learning |
| CN110097974A (en) * | 2019-05-15 | 2019-08-06 | 天津医科大学肿瘤医院 | A kind of nasopharyngeal carcinoma far-end transfer forecasting system based on deep learning algorithm |
| CN110378885A (en) * | 2019-07-19 | 2019-10-25 | 王晓骁 | A kind of focal area WSI automatic marking method and system based on machine learning |
| CN110459300A (en) * | 2019-07-22 | 2019-11-15 | 中国石油大学(华东) | A kind of lung cancer pathology type diagnostic method based on computer vision and CT images |
| CN111340128A (en) * | 2020-03-05 | 2020-06-26 | 上海市肺科医院（上海市职业病防治院） | Lung cancer metastatic lymph node pathological image recognition system and method |
| CN111369573A (en) * | 2020-03-05 | 2020-07-03 | 上海市肺科医院（上海市职业病防治院） | Auxiliary identification system and method for lung adenocarcinoma subtypes |
| CN111461243A (en) * | 2020-04-08 | 2020-07-28 | 中国医学科学院肿瘤医院 | Classification method, classification device, electronic equipment and computer-readable storage medium |
| CN112241945A (en) * | 2019-07-17 | 2021-01-19 | 广州信瑞医疗技术有限公司 | Digital pathological image intelligent analysis method with deep learning algorithm and hardware integral optimization |
Family Cites Families (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6678669B2 (en) * | 1996-02-09 | 2004-01-13 | Adeza Biomedical Corporation | Method for selecting medical and biochemical diagnostic tests using neural network-related applications |
| US8077958B2 (en) * | 2006-06-30 | 2011-12-13 | University Of South Florida | Computer-aided pathological diagnosis system |
| CN107028593A (en) * | 2017-04-14 | 2017-08-11 | 成都知识视觉科技有限公司 | A kind of aided detection method of breast ductal carcinoma in situ |
| IL313642A (en) * | 2019-03-08 | 2024-08-01 | Univ Southern California | Improved histopathology classification through machine self-learning of " tissue fingerprints" |
| AU2020274091A1 (en) * | 2019-05-14 | 2021-12-09 | Tempus Ai, Inc. | Systems and methods for multi-label cancer classification |
| CN110763678A (en) * | 2019-09-12 | 2020-02-07 | 杭州迪英加科技有限公司 | Pathological section interpretation method and system |
| CN110727097B (en) * | 2019-12-19 | 2020-05-05 | 上海兰脉信息科技有限公司 | Pathological microscopic image real-time acquisition and analysis system, method, device and medium |
| CN111370128A (en) * | 2020-03-05 | 2020-07-03 | 上海市肺科医院（上海市职业病防治院） | Lung cancer patient prognosis prediction system and method |
| CN111967465A (en) * | 2020-07-07 | 2020-11-20 | 广州金域医学检验中心有限公司 | Method, system, computer device and storage medium for evaluating tumor cell content |
| CN111986150B (en) * | 2020-07-17 | 2024-02-09 | 万达信息股份有限公司 | The method comprises the following steps of: digital number pathological image Interactive annotation refining method |
-
2021
- 2021-02-02 CN CN202110144607.7A patent/CN112884724B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109754879A (en) * | 2019-01-04 | 2019-05-14 | 湖南兰茜生物科技有限公司 | A kind of lung cancer computer aided detection method and system based on deep learning |
| CN110097974A (en) * | 2019-05-15 | 2019-08-06 | 天津医科大学肿瘤医院 | A kind of nasopharyngeal carcinoma far-end transfer forecasting system based on deep learning algorithm |
| CN112241945A (en) * | 2019-07-17 | 2021-01-19 | 广州信瑞医疗技术有限公司 | Digital pathological image intelligent analysis method with deep learning algorithm and hardware integral optimization |
| CN110378885A (en) * | 2019-07-19 | 2019-10-25 | 王晓骁 | A kind of focal area WSI automatic marking method and system based on machine learning |
| CN110459300A (en) * | 2019-07-22 | 2019-11-15 | 中国石油大学(华东) | A kind of lung cancer pathology type diagnostic method based on computer vision and CT images |
| CN111340128A (en) * | 2020-03-05 | 2020-06-26 | 上海市肺科医院（上海市职业病防治院） | Lung cancer metastatic lymph node pathological image recognition system and method |
| CN111369573A (en) * | 2020-03-05 | 2020-07-03 | 上海市肺科医院（上海市职业病防治院） | Auxiliary identification system and method for lung adenocarcinoma subtypes |
| CN111461243A (en) * | 2020-04-08 | 2020-07-28 | 中国医学科学院肿瘤医院 | Classification method, classification device, electronic equipment and computer-readable storage medium |
Non-Patent Citations (1)
| Title |
|---|
| 基于卷积神经网络的肺癌病理图像分类;宁静艳等;《软件导刊》;20190215;第18卷(第2期);第141-144页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112884724A (en) | 2021-06-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112884724B (en) | Intelligent judgment method and system for lung cancer histopathological typing | |
| US11164312B2 (en) | System and method to quantify tumor-infiltrating lymphocytes (TILs) for clinical pathology analysis based on prediction, spatial analysis, molecular correlation, and reconstruction of TIL information identified in digitized tissue images | |
| CN112101451B (en) | Breast cancer tissue pathological type classification method based on generation of antagonism network screening image block | |
| CN107280697A (en) | Lung neoplasm grading determination method and system based on deep learning and data fusion | |
| CN107330263A (en) | A kind of method of area of computer aided breast invasive ductal carcinoma histological grading | |
| CN110991536B (en) | Training method of early warning model of primary liver cancer | |
| CN110796661B (en) | Fungal microscopic image segmentation detection method and system based on convolutional neural network | |
| CN109124660B (en) | Gastrointestinal stromal tumor postoperative risk detection method and system based on deep learning | |
| CN113793305A (en) | Pathological image classification and identification method and system integrating multiple information | |
| CN116309368B (en) | Lung cancer pathological diagnosis system based on deep migration learning | |
| CN113628199B (en) | Pathological picture stained tissue area detection method, pathological picture stained tissue area detection system and prognosis state analysis system | |
| US20230360208A1 (en) | Training end-to-end weakly supervised networks at the specimen (supra-image) level | |
| CN113657449A (en) | Traditional Chinese medicine tongue picture greasy classification method containing noise labeling data | |
| CN116978543A (en) | Artificial intelligent auxiliary marrow tumor pathological diagnosis device | |
| CN116664523A (en) | Method for constructing renal clear cell carcinoma grading model based on deep learning | |
| CN113435469A (en) | Kidney tumor enhanced CT image automatic identification system based on deep learning and training method thereof | |
| CN116468690A (en) | Subtype analysis system of invasive non-mucous lung adenocarcinoma based on deep learning | |
| CN115564750A (en) | Intraoperative frozen slice image identification method, intraoperative frozen slice image identification device, intraoperative frozen slice image identification equipment and intraoperative frozen slice image storage medium | |
| CN111312372A (en) | Method and device for establishing medical image standard test data set | |
| CN116469513B (en) | Colorectal cancer individuation treatment and prognosis information prediction system based on deep learning | |
| WO2023240819A1 (en) | Pathological analysis module for thyroid disease | |
| Lu et al. | AI-Based Computational Pathology and Its Contribution to Precision Medicine | |
| Kumar et al. | Innovative Breast Cancer Detection Through Mammogram Analysis | |
| CN117275679A (en) | Pathological image-based heat shock path analysis method | |
| Li et al. | Deep Learning based Intelligent Tumor Analytics Framework for Quantitative Grading and Analyzing Cancer Metastasis: Case of Lymph Node Breast Cancer |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |