CN113762395A - Pancreatic bile duct type ampulla carcinoma classification model generation method and image classification method - Google Patents
Pancreatic bile duct type ampulla carcinoma classification model generation method and image classification method Download PDFInfo
- Publication number
- CN113762395A CN113762395A CN202111056391.5A CN202111056391A CN113762395A CN 113762395 A CN113762395 A CN 113762395A CN 202111056391 A CN202111056391 A CN 202111056391A CN 113762395 A CN113762395 A CN 113762395A
- Authority
- CN
- China
- Prior art keywords
- image
- classification model
- ampulla
- carcinoma
- cancer
- 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.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/2431—Multiple classes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
Abstract
The invention discloses a method for generating a classification model of cholangiopancreaticobiliary type ampulla carcinoma and an image classification method, wherein the model generation method comprises the following steps: constructing an initial classification model, wherein the initial classification model comprises an image preprocessing module, a cell segmentation module, a cell morphological feature extraction module and a classification module; marking the collected pathological sections of the bile duct cancer and the pancreatic cancer to form a digital pathological image marking library; dividing the digital pathological image labeling library into training set data and test set data according to a preset proportion, wherein the training set data comprises bile duct cancer pathological sections and pancreatic cancer pathological sections which are subjected to labeling processing; and training the initial classification model by adopting training set data, and completing parameter adjustment of the initial classification model to obtain the pancreaticobiliary duct type ampulla carcinoma classification model. The invention firstly constructs a classification model to classify the pancreas and bile duct type periampulla carcinoma according to the H & E dyed full-section digital pathological image, namely, whether the tumor originates from pancreatic duct (pancreatic cancer) or bile duct (bile duct cancer) is judged.
Description
Technical Field
The invention relates to the technical field of image recognition, in particular to a pancreaticobiliary duct type ampulla carcinoma classification model generation method and an image classification method.
Background
Periampulla Carcinoma (VPC) refers to a tumor of ampulla of vater, lower end of common bile duct, opening of pancreatic duct, duodenal papilla, and the duodenal mucosa of adnexa. The ampulla is specifically classified into carcinoma of head of pancreas, common bile duct carcinoma of pancreas, ampulla carcinoma and duodenal carcinoma, as shown in fig. 1, and the latter three may be collectively referred to as ampulla carcinoma. The survival rates of periampulla cancers, which are traditionally classified according to their location of anatomical origin, i.e., duodenum, ampulla, distal common bile duct, or head of pancreas, vary widely. However, they can be histopathologically alternatively subdivided into intestinal or pancreaticobiliary types, which can more accurately estimate prognosis. Due to the wide variation in clinical outcomes and treatment management, accurate differential diagnosis of the tissue origin of pancreaticobiliary peri-ampulla cancer is important. For example, although the tumor positions at the distal common bile duct and pancreatic duct opening are adjacent, the natural course of disease, the surgical resection rate and the prognosis are different, and they should be treated differently in the specific clinical diagnosis and treatment process. Therefore, the early diagnosis of the periampulla carcinoma and the accurate judgment of the origin of the tumor are of great significance for improving the prognosis of the periampulla carcinoma.
According to the combination of pathological histological morphological characteristics and immunohistochemistry, the periampulla ampullata cancer can be divided into an intestinal type and a pancreatic-biliary type, but because an effective immunohistochemical marker is lacked, the further subtype analysis of the pancreatic-biliary type cannot be carried out clinically at present, so that an effective and reliable method for identifying and diagnosing the tissue origin of the periampulla ampullata cancer of the pancreatic-biliary type is urgently needed clinically, the tissue origin of a tumor is judged according to biological characteristics, and an optimal treatment scheme is specified.
Accordingly, the prior art is yet to be improved and developed.
Disclosure of Invention
In view of the above-mentioned shortcomings of the prior art, an object of the present invention is to provide a method for generating a classification model of pancreaticobiliary ampulla cancer and a method for classifying images, which aims to solve the problem that the prior art cannot automatically classify pathological sections of pancreaticobiliary ampulla cancer into subtypes further.
The technical scheme of the invention is as follows:
a method for generating a classification model of cholangiopancreaticobiliary type ampulla cancer comprises the following steps:
constructing an initial classification model, wherein the initial classification model comprises an image preprocessing module, a cell segmentation module, a cell morphological feature extraction module and a classification module;
marking the collected pathological sections of the bile duct cancer and the pancreatic cancer to form a digital pathological image marking library;
dividing the digital pathological image labeling library into training set data and test set data according to a preset proportion, wherein the training set data comprises bile duct cancer pathological sections and pancreatic cancer pathological sections which are subjected to labeling processing;
and training the initial classification model by adopting the training set data, finishing parameter adjustment of the initial classification model, and obtaining the pancreaticobiliary duct type ampulla carcinoma classification model.
The pancreaticobiliary duct type ampulla carcinoma classification model generation method comprises the steps that the image preprocessing module is used for combining a dyeing density graph of an image in the digital pathological image labeling library with a dyeing color basis of a preselected target image, so that the color of the image in the digital pathological image labeling library is changed, the structure of the image in the digital pathological image labeling library is reserved, color normalization processing of the image in the digital pathological image labeling library is achieved, and a color normalized image is obtained.
The method for generating the pancreaticobiliary duct type ampulla carcinoma classification model comprises the following steps of dividing the color normalized image into small image blocks, and dividing cell nucleuses in the small image blocks.
The method for generating the classification model of the cholangiopancreatic ampulla carcinoma comprises a cell morphology feature extraction module, a cell classification module and a classification module, wherein the cell morphology feature extraction module is used for extracting 10 types of cell nucleus horizontal features of the cell nucleus and converting the cell nucleus horizontal features into image horizontal features.
The method for generating the classification model of cholangiopancreatic ampulla carcinoma comprises the following steps of generating 10 types of horizontal characteristics of cell nuclei, wherein the horizontal characteristics of the cell nuclei comprise the area of the cell nuclei, the length of the long axis of the cell nuclei, the length of the short axis of the cell nuclei, the ratio of the length of the long axis to the length of the short axis of the cell nuclei, the average pixel value of the R channel of the cell nuclei, the average pixel value of the G channel of the cell nuclei, the average pixel value of the B channel of the cell nuclei, and the average value, the maximum value and the minimum value of the distance from the current cell nuclei to the surrounding adjacent cells in a Delaunay triangularization map.
The method for generating the classification model of the cholangiopancreatic ampulla carcinoma comprises the steps that the image horizontal characteristics comprise 10 histogram characteristics and 5 distribution statistic characteristics.
The classification module is used for screening the extracted image horizontal characteristics and further classifying the cholangiopancreatico type ampulla carcinoma of the images in the digital pathological image labeling library according to the screened image horizontal characteristics.
An image classification method, comprising the steps of:
acquiring a pathological section image to be classified;
and inputting the pathological section image to be classified into the classification model constructed by the pancreaticobiliary duct type ampulla cancer classification model generating method, and outputting a classification result.
A storage medium storing one or more programs executable by one or more processors to perform the steps of the cholangiopancreatic carcinoma classification model generation method of the present invention or the steps of the image classification method of the present invention.
A terminal device, comprising: a processor, a memory, and a communication bus; the memory has stored thereon a computer readable program executable by the processor; the communication bus realizes connection communication between the processor and the memory; the processor, when executing the computer readable program, implements the steps of the method for generating a classification model of cholangiopancreaticobiliary carcinoma of the present invention or the steps of the method for image classification of the present invention.
Has the advantages that: the invention provides a method for generating a classification model of cholangiopancreaticobiliary type ampulla carcinoma, which comprises the steps of firstly establishing a digital pathological image labeling library, carrying out structure-keeping color normalization processing on image data in the digital pathological image labeling library, then extracting cell morphological characteristics, using the extracted characteristics in an initial classification model, and obtaining the classification model of cholangiopancreaticobiliary type ampulla carcinoma through a series of parameter adjustment. The H & E stained full-section digital pathological image is used for constructing the pancreaticobiliary duct type ampulla cancer classification model for the first time to classify the subtype of the pancreaticobiliary duct type ampulla cancer section image, and the result can help doctors to make a better treatment scheme for patients.
Drawings
Fig. 1 is a schematic view of the ampulla position.
FIG. 2 is a flowchart of a method for generating a classification model of cholangiopancreaticosa type ampulla carcinoma according to the present invention.
FIG. 3 is a block diagram of the image classification method according to the present invention.
FIG. 4 is a color normalization diagram of the images in the digital pathology image labeling library with the same staining condition.
FIG. 5 is a schematic illustration of the segmentation of nuclei according to the present invention.
FIG. 6 is a schematic representation of nuclear-scale features.
FIG. 7 is a diagram of image level features of cell nuclei.
Fig. 8 is a schematic block diagram of a terminal device according to the present invention.
Detailed Description
The invention provides a pancreaticobiliary duct type ampulla carcinoma classification model generation method and an image classification method, and the invention is further described in detail below in order to make the purpose, technical scheme and effect of the invention clearer and clearer. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
At present, there are methods for differential diagnosis of cancer around ampulla by using CT energy spectrum imaging, Magnetic Resonance (MR) image, morphology, immunohistochemistry, multi-layer spiral CT, magnetic resonance pancreaticocholangiography, quantitative analysis and imaging techniques. Most of the methods only classify the periampulla carcinoma according to the anatomical origin position by the traditional classification method, and only use the imaging images (CT, MRI, etc.) without the knowledge of the morphological characteristics of the pathological images. Although a small part of methods combines pathological image morphological feature knowledge and immunohistochemistry, only the intestinal type and the pancreaticobiliary type in the cancer around the ampulla are classified, and the pancreaticobiliary subtype is not further analyzed.
In pathology, the birth of a digital slice opens the prelude to digital pathology. After the pathological image is digitized, the existing artificial intelligence and machine learning algorithms can be applied or new algorithms can be developed, and the algorithms can help pathologists to complete repeated and boring low-level tasks such as cell counting and mitotic cell detection, so that the workload of the pathologists is reduced; a large number of quantitative features can also be extracted from pathological images to accomplish some high-level tasks, such as automatic diagnosis and prognosis prediction of cancer. After the characteristics relevant to differential diagnosis of the periampulla cancer are screened from the extracted large number of image characteristics, a classifier for judging the periampulla cancer subtype can be obtained by training a machine learning relevant model, so that a doctor can be helped to accurately diagnose, the clinical diagnosis and treatment effect of the periampulla cancer is improved, and the prognosis of a patient is improved.
Based on this, the invention provides a method for generating a classification model of cholangiopancreaticobiliary type ampulla cancer, as shown in fig. 2, which comprises the following steps:
s10, constructing an initial classification model, wherein the initial classification model comprises an image preprocessing module, a cell segmentation module, a cell morphological feature extraction module and a classification module;
s20, marking the collected pathological sections of the bile duct cancer and the pancreatic cancer to form a digital pathological image marking library;
s30, dividing the digital pathological image labeling library into training set data and test set data according to a preset proportion, wherein the training set data comprise bile duct cancer pathological sections and pancreatic cancer pathological sections which are subjected to labeling processing;
and S40, training the initial classification model by adopting the training set data, completing parameter adjustment of the initial classification model, and obtaining the pancreaticobiliary duct type ampulla carcinoma classification model.
In the embodiment, a digital pathological image labeling library is established, after structure-preserving color normalization processing is performed on image data in the digital pathological image labeling library, morphological characteristics of cells are extracted, the extracted characteristics are used for an initial classification model, and the pancreaticobiliary duct type ampulla carcinoma classification model is obtained through a series of parameter adjustment. In the embodiment, a classification model is constructed to classify the subtype of the pancreaticobiliary ampulla carcinoma section image based on the H & E stained full-section digital pathological image for the first time, and the result can help a doctor to make a better treatment scheme for a patient.
The method for generating the classification model of cholangiopancreaticobiliary type ampulla is further explained by the following specific embodiment:
the invention aims to realize subtype classification of pancreaticobiliary type pathological section images of the carcinoma around the ampulla, the carcinoma around the ampulla is clinically classified into an intestinal type and a pancreaticobiliary type, the intestinal type can be distinguished through an immunohistochemical marker at present, but the pancreaticobiliary type cannot be further subjected to subgroup analysis clinically at present because of lack of an effective immunohistochemical marker. Therefore, the present invention is directed to differentiating the pancreaticobiliary type in the periampulla cancer, and since there is no gold standard for classification of the pancreaticobiliary type at present, the present invention requires the use of pathological sections of ductal adenocarcinoma in pancreas (i.e., pathological sections of pancreatic cancer) and pathological sections of cell cancer of bile duct in liver (i.e., pathological sections of bile duct cancer) to train an initial classification model, so that the two data are selected because the ductal adenocarcinoma in pancreas and the bile duct in liver are far from the ampulla, and are clearly diagnosed as pancreatic cancer or bile duct cancer. The development of the invention is therefore based on three sets of data collected, namely biliary duct cancer (38 cases) and pancreatic cancer (38 cases) digital pathology images downloaded from a cancer genomic map (TCGA) website, the data sets each containing full-section digital pathology images and clinical pathology grading, staging and the like data; data two is pathological sections of patients with cholangiocarcinoma (71 cases) and pancreatic carcinoma (70 cases) and periampulla carcinoma (35 cases) collected in order from the affiliated tumor hospital of Zhongshan university; data three are pathological sections of patients with cholangiocarcinoma (16 cases) and pancreatic carcinoma (30 cases) and periampulla carcinoma (10 cases) collected from the state secondary hospital in Ningbo and the northern regional people hospital, Zhejiang province, northern prefecture. After the data collection is finished, the collected pathological sections are scanned into an electronic computer by a section scanner, and the digital image matrix is stored by RGB three channels.
After the data acquisition is finished, screening digital pathological sections with pathological expert diagnosis information from the acquired data to establish a digital pathological image database, and carrying out strict tumor region labeling on all the data pathological sections of the database to form a digital pathological image labeling database.
After a digital pathological image labeling library is constructed, dividing the digital pathological image labeling library into training set data and test set data according to a preset proportion, wherein the training set data comprises bile duct cancer pathological sections and pancreatic cancer pathological sections which are subjected to labeling processing; and training an initial classification model by adopting the training set data, completing parameter adjustment of the initial classification model, and obtaining the pancreaticobiliary duct type ampulla cancer classification model, wherein the initial classification model comprises an image preprocessing module, a cell segmentation module, a cell morphological characteristic extraction module and a classification module.
In this embodiment, as shown in fig. 3, the image preprocessing module is configured to combine a staining density map of an image in the digital pathology image labeling library with a staining color basis of a preselected target image, so as to change a color of the image in the digital pathology image labeling library and retain a structure of the image in the digital pathology image labeling library, thereby implementing color normalization processing on the image in the digital pathology image labeling library to obtain a color normalized image. In particular, staining and scanning of tissue samples for microscopic examination is fraught with undesirable color variations due to differences in the raw materials and manufacturing techniques of the staining suppliers, the staining protocols of the laboratory, and the color response of the digital scanners. Color normalization of tissue images is helpful to both pathologists and software when comparing tissue samples. Techniques for natural imaging cannot take advantage of the structural properties of stained tissue samples and produce undesirable color distortions. To mitigate the effect of color differences between different images on the feature extraction and model training process, this embodiment uses the color normalization method proposed by Vahadane et al, which defines the physical phenomenon of tissue structure by decomposing the images into sparse and non-negative staining density maps in an unsupervised manner. As shown in fig. 4, for a given image, its stain density map is combined with the basis of the stain color of the pathologist's preferred target image, thereby changing only its color while preserving the structure described in its original image.
In this embodiment, the pathological image feature extraction process includes 3 steps: segmentation of cell nuclei, feature extraction at the cell nucleus level, and feature extraction at the image level. Wherein, as shown in fig. 3, the cell segmentation module is configured to segment the color normalized image into image patches and segment nuclei in the image patches. Specifically, for the segmentation of the cell nuclei, the method proposed by Phoulady et al, which is an unsupervised method for segmenting the cell nuclei in the histopathological image, is adopted in this embodiment. Specifically, after the preprocessing steps of color deconvolution and image reconstruction, segmentation operation is carried out, the segmentation step comprises a multi-level threshold value and a series of morphological operations, and parameters can be set adaptively according to the content of an image, so that the method does not need parameter learning or training data and is insensitive to the change of the dyeing intensity. As shown in FIG. 5, the present invention uses this method to segment the nuclei in the image after the full-slide image is cut into image patches.
In the present embodiment, as shown in fig. 3, the cellular morphological feature extraction module is configured to extract 10 types of cell nucleus level features of the cell nucleus and convert the cell nucleus level features into image level features. Specifically, for each of the divided nuclei, 10 types of characteristics of the nucleus level are extracted, as shown in fig. 6, the 10 types of characteristics of the nucleus level include the area of the nucleus, the length of the long axis of the nucleus, the length of the short axis of the nucleus, the ratio of the length of the long axis and the short axis of the nucleus, the average pixel value of the R channel of the nucleus, the average pixel value of the G channel of the nucleus, the average pixel value of the B channel of the nucleus, and the average, maximum, and minimum values of the distances from the current nucleus to the surrounding neighboring cells in a Delaunay triangulation map, which is constructed based on the positions of the nuclei, in which each nucleus is a node and is connected to the surrounding nuclei.
After the feature extraction at the cell level is completed, converting the feature of each type of cell nucleus level into the feature at the image level, wherein the feature at the image level comprises 10 histogram features and 5 distribution statistic features, and the 5 distribution statistic features are respectively a mean value, a standard deviation, a skewness, a kurtosis and an entropy. In this embodiment, each feature at the cell nucleus level corresponds to a feature at the 15 image levels, so that each patient has a total of 150 pathological image features. As shown in fig. 7, taking the area of the cell nucleus as an example, the names of the histogram features of the corresponding 10 image levels are: area _ bin1, area _ bin2, area _ bin3, …, area _ bin10, while area _ bin1 represents the percentage of very small cells in the pathology image, area _ bin10 represents the percentage of very large cells in the pathology image, the names of image level features corresponding to other cell levels, and so on. Taking the area of the cell nucleus as an example, the names of the distribution statistics of the corresponding 5 image levels are respectively: area _ mean, area _ SD, area _ skewness, area _ kurtosis, and area _ entry.
In this embodiment, the classification module is configured to screen the extracted image horizontal features and classify pancreaticobiliary ampulla carcinoma of the images in the digital pathological image labeling library according to the screened image horizontal features. Specifically, because information carried by different types of features may have redundancy, it is necessary to select features from high-dimensional features that can effectively predict the cancer subtype around the ampulla to reduce the influence caused by irrelevant features, which not only can avoid overfitting or dimensionality disasters, reduce the complexity of calculation, but also can effectively improve the prediction performance of the model. In order to select image features relevant to differential diagnosis of periampulla carcinoma, the invention uses the skearn library feature _ selection feature selection algorithm SelectKBest (score _ func, k) to screen features: where score _ func is passed in as a function to calculate the score, F _ classif by default, which calculates the analysis of variance F-value between the univariate and the training target (Anova F-value); k afferents the number of variables we want to leave from high to low according to the score, the default is 10, the appropriate k value is set, and the pathological image features which are finally used for training the initial classification model are screened out by the method.
In some embodiments, the initial classification model is trained using the screened pathological image features to derive a classification model for predicting pancreaticobiliary ampulla carcinoma. Specifically, the feature data extracted from the first embodiment data is randomly divided into a training set (80%) and a test set (20%), appropriate model parameters are selected through five-fold cross validation on the training set, after the parameters are determined, the selected logistic regression, random forest, support vector machine, K-neighborhood model and the like are trained by the training set, a model which is most suitable for the task of the invention is selected, and the performance of the classification model is tested on the test set. The characteristic data of bile duct cancer and pancreatic cancer of the second data and the third data of the embodiment are taken as two independent external verification sets to verify the generalization performance of the model. And further optimizing parameters of the test result to obtain a final pancreaticobiliary ampulla cancer classification model, wherein the pancreaticobiliary ampulla cancer classification model can be used for predicting whether the tumor of the patient with the periampulla cancer originates from a pancreatic duct or a bile duct. And inputting the characteristic data of the periampulla carcinoma of the data II and the data III into the pancreaticobiliary duct type classification model to obtain a prediction result, wherein the prediction result can help a doctor to make a better treatment scheme for a patient.
Based on the classification model of cholangiopancreaticosa type ampulla carcinoma generated by the above embodiment, the present invention also provides an image classification method, which specifically includes the steps of: acquiring a pathological section image to be classified; and inputting the pathological section image to be classified into the pancreaticobiliary duct type ampulla cancer classification model, and outputting a predicted classification result. The invention is based on H & E stained full-section digital pathological images for the first time, and a machine learning model is constructed to identify and classify the tissue origin of the pancreaticobiliary duct type ampulla carcinoma.
In some embodiments, a storage medium is also provided, wherein the storage medium stores one or more programs executable by one or more processors to implement the steps of the pancreaticobiliary ampulla carcinoma classification model generation method of the present invention or the image classification method of the present invention.
The present application also provides a terminal device, as shown in fig. 8, which includes at least one processor (processor) 20; a display screen 21; and a memory (memory)22, and may further include a communication Interface (Communications Interface)23 and a bus 24. The processor 20, the display 21, the memory 22 and the communication interface 23 can communicate with each other through the bus 24. The display screen 21 is configured to display a user guidance interface preset in the initial setting mode. The communication interface 23 may transmit information. The processor 20 may invoke logic instructions in the memory 22 to perform the steps of the pancreaticobiliary type ampulla carcinoma classification model generation method in the above-described embodiments or the image classification method of the present invention.
Furthermore, the logic instructions in the memory 22 may be implemented in software functional units and stored in a computer readable storage medium when sold or used as a stand-alone product.
The memory 22, which is a computer-readable storage medium, may be configured to store a software program, a computer-executable program, such as program instructions or modules corresponding to the methods in the embodiments of the present disclosure. The processor 20 executes the functional application and data processing, i.e. implements the method in the above-described embodiments, by executing the software program, instructions or modules stored in the memory 22.
The memory 22 may include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required for at least one function; the storage data area may store data created according to the use of the terminal device, and the like. Further, the memory 22 may include a high speed random access memory and may also include a non-volatile memory. For example, a variety of media that can store program codes, such as a usb disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk, may also be transient storage media.
In addition, the specific processes loaded and executed by the storage medium and the instruction processors in the terminal device are described in detail in the above method, and are not stated herein.
It is to be understood that the invention is not limited to the examples described above, but that modifications and variations may be effected thereto by those of ordinary skill in the art in light of the foregoing description, and that all such modifications and variations are intended to be within the scope of the invention as defined by the appended claims.
Claims (10)
1. A method for generating a classification model of cholangiopancreaticobiliary type ampulla cancer is characterized by comprising the following steps:
constructing an initial classification model, wherein the initial classification model comprises an image preprocessing module, a cell segmentation module, a cell morphological feature extraction module and a classification module;
marking the collected pathological sections of the bile duct cancer and the pancreatic cancer to form a digital pathological image marking library;
dividing the digital pathological image labeling library into training set data and test set data according to a preset proportion, wherein the training set data comprises bile duct cancer pathological sections and pancreatic cancer pathological sections which are subjected to labeling processing;
and training the initial classification model by adopting the training set data, finishing parameter adjustment of the initial classification model, and obtaining the pancreaticobiliary duct type ampulla carcinoma classification model.
2. The method for generating a classification model of cholangiopancreatic ampulla carcinoma according to claim 1, wherein the image preprocessing module is configured to combine a staining density map of an image in the digital pathological image labeling library with a staining color basis of a preselected target image, thereby changing a color of the image in the digital pathological image labeling library and preserving a structure of the image in the digital pathological image labeling library, so as to perform color normalization processing on the image in the digital pathological image labeling library, thereby obtaining a color-normalized image.
3. The method of generating a classification model for cholangiopancreatic ampulla carcinoma according to claim 2, wherein the cell segmentation module is configured to segment the color-normalized image into image patches and segment nuclei in the image patches.
4. The method of claim 3, wherein the cellular morphological feature extraction module is configured to extract 10 types of cell nucleus level features of the cell nucleus and convert the cell nucleus level features into image level features.
5. The pancreaticobiliary ampulla carcinoma classification model generation method according to claim 4, wherein the 10-type cell nucleus horizontal features include an area of a cell nucleus, a length of a long axis of the cell nucleus, a length of a short axis of the cell nucleus, a ratio of the lengths of the long axis and the short axis of the cell nucleus, an average pixel value of an R channel of the cell nucleus, an average pixel value of a G channel of the cell nucleus, an average pixel value of a B channel of the cell nucleus, and an average value, a maximum value, and a minimum value of a distance from a current cell nucleus to surrounding neighboring cells in a Delaunay triangulated map.
6. The method of generating a classification model for cholangiopancreatic ampulla carcinoma according to claim 5, wherein the image level features include 10 histogram features and 5 distribution statistic features.
7. The method for generating a classification model of cholangiopancreatic ampulla cancer according to claim 6, wherein the classification module is used for screening the extracted image horizontal features and classifying cholangiopancreatic ampulla cancer of images in the digital pathological image labeling library according to the screened image horizontal features.
8. An image classification method, characterized by comprising the steps of:
acquiring a pathological section image to be classified;
inputting the pathological section image to be classified into the classification model constructed by the pancreaticobiliary duct type ampulla cancer classification model generating method according to any one of claims 1 to 7, and outputting a classification result.
9. A storage medium storing one or more programs executable by one or more processors to perform the steps of the method for generating a classification model of cholangiopancreatic ampulla carcinoma according to any one of claims 1 to 7 or the steps of the method for image classification according to claim 8.
10. A terminal device, comprising: a processor, a memory, and a communication bus; the memory has stored thereon a computer readable program executable by the processor; the communication bus realizes connection communication between the processor and the memory; the processor, when executing the computer readable program, carries out the steps of the method for generating a classification model for cholangiopancreaticobiliary carcinoma as defined in any one of claims 1 to 7 or the steps of the method for image classification as defined in claim 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111056391.5A CN113762395B (en) | 2021-09-09 | 2021-09-09 | Pancreatic bile duct type ampulla carcinoma classification model generation method and image classification method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111056391.5A CN113762395B (en) | 2021-09-09 | 2021-09-09 | Pancreatic bile duct type ampulla carcinoma classification model generation method and image classification method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113762395A true CN113762395A (en) | 2021-12-07 |
| CN113762395B CN113762395B (en) | 2022-08-19 |
Family
ID=78794437
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111056391.5A Active CN113762395B (en) | 2021-09-09 | 2021-09-09 | Pancreatic bile duct type ampulla carcinoma classification model generation method and image classification method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113762395B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114693662A (en) * | 2022-04-12 | 2022-07-01 | 吉林大学 | Histopathological section subtype classification method and device, medium and terminal |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017087415A1 (en) * | 2015-11-17 | 2017-05-26 | The Board Of Trustees Of The Leland Stanford Junior University | Profiling of pathology images for clinical applications |
| CN106874687A (en) * | 2017-03-03 | 2017-06-20 | 深圳大学 | Pathological section image intelligent sorting technique and device |
| CN108364288A (en) * | 2018-03-01 | 2018-08-03 | 北京航空航天大学 | Dividing method and device for breast cancer pathological image |
| CN109886922A (en) * | 2019-01-17 | 2019-06-14 | 丽水市中心医院 | Hepatocellular carcinoma automatic grading method based on SE-DenseNet deep learning frame and multi-modal Enhanced MR image |
| 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 |
| CN111402267A (en) * | 2020-03-13 | 2020-07-10 | 中山大学孙逸仙纪念医院 | Segmentation method, device and terminal for epithelial cell nucleus in prostate cancer pathological image |
| CN112435243A (en) * | 2020-11-26 | 2021-03-02 | 山东第一医科大学附属省立医院(山东省立医院) | Automatic analysis system and method for full-slice digital pathological image |
| CN113052806A (en) * | 2021-03-15 | 2021-06-29 | 黑龙江机智通智能科技有限公司 | Canceration degree grading system |
-
2021
- 2021-09-09 CN CN202111056391.5A patent/CN113762395B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017087415A1 (en) * | 2015-11-17 | 2017-05-26 | The Board Of Trustees Of The Leland Stanford Junior University | Profiling of pathology images for clinical applications |
| CN106874687A (en) * | 2017-03-03 | 2017-06-20 | 深圳大学 | Pathological section image intelligent sorting technique and device |
| CN108364288A (en) * | 2018-03-01 | 2018-08-03 | 北京航空航天大学 | Dividing method and device for breast cancer pathological image |
| CN109886922A (en) * | 2019-01-17 | 2019-06-14 | 丽水市中心医院 | Hepatocellular carcinoma automatic grading method based on SE-DenseNet deep learning frame and multi-modal Enhanced MR image |
| 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 |
| CN111402267A (en) * | 2020-03-13 | 2020-07-10 | 中山大学孙逸仙纪念医院 | Segmentation method, device and terminal for epithelial cell nucleus in prostate cancer pathological image |
| CN112435243A (en) * | 2020-11-26 | 2021-03-02 | 山东第一医科大学附属省立医院(山东省立医院) | Automatic analysis system and method for full-slice digital pathological image |
| CN113052806A (en) * | 2021-03-15 | 2021-06-29 | 黑龙江机智通智能科技有限公司 | Canceration degree grading system |
Non-Patent Citations (2)
| Title |
|---|
| JOHANNES R.K.等: "Incorporation of a Molecular Prognostic Classifier Improves Conventional Non–Small Cell Lung Cancer Staging", 《JOURNAL OF THORACIC ONCOLOGY》 * |
| 戴少明: "多层螺旋CT联合MRI在壶腹周围癌术前诊断中的应用价值及影像学特点研究", 《中外医学研究》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114693662A (en) * | 2022-04-12 | 2022-07-01 | 吉林大学 | Histopathological section subtype classification method and device, medium and terminal |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113762395B (en) | 2022-08-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11842556B2 (en) | Image analysis method, apparatus, program, and learned deep learning algorithm | |
| Cruz-Roa et al. | Accurate and reproducible invasive breast cancer detection in whole-slide images: A Deep Learning approach for quantifying tumor extent | |
| US10783641B2 (en) | Systems and methods for adaptive histopathology image unmixing | |
| CN109903284B (en) | HER2 immunohistochemical image automatic discrimination method and system | |
| Mi et al. | Deep learning-based multi-class classification of breast digital pathology images | |
| US10861156B2 (en) | Quality control for digital pathology slides | |
| Xu et al. | Deep learning for histopathological image analysis: Towards computerized diagnosis on cancers | |
| CN111402267B (en) | Segmentation method, device and terminal of epithelial cell nuclei in prostate cancer pathological image | |
| CN111462042A (en) | Cancer prognosis analysis method and system | |
| Goceri et al. | Quantitative validation of anti‐PTBP1 antibody for diagnostic neuropathology use: Image analysis approach | |
| CN112215790A (en) | KI67 index analysis method based on deep learning | |
| Ström et al. | Pathologist-level grading of prostate biopsies with artificial intelligence | |
| CN112183557A (en) | MSI prediction model construction method based on gastric cancer histopathology image texture features | |
| CN107567631B (en) | Tissue sample analysis techniques | |
| Chatterjee et al. | A novel method for IDC prediction in breast cancer histopathology images using deep residual neural networks | |
| CN113570619A (en) | Computer-aided pancreas pathology image diagnosis system based on artificial intelligence | |
| CN114841947A (en) | Method and device for multi-scale feature extraction and prognosis analysis of H & E staining pathological image tumor region | |
| CN113762395B (en) | Pancreatic bile duct type ampulla carcinoma classification model generation method and image classification method | |
| Kloeckner et al. | Multi-categorical classification using deep learning applied to the diagnosis of gastric cancer | |
| Lal et al. | A robust method for nuclei segmentation of H&E stained histopathology images | |
| Huang et al. | HEp-2 cell images classification based on textural and statistic features using self-organizing map | |
| Chauhan et al. | Exploring genetic-histologic relationships in breast cancer | |
| KR101297336B1 (en) | Device For Classifying Mucinous Cystadenoma | |
| CN114648509B (en) | Thyroid cancer detection system based on multi-classification task | |
| CN115205588A (en) | Bladder cancer muscle layer infiltration pathology auxiliary diagnosis method and device based on artificial intelligence |
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 |