Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Breast cancer histopathological type classification method based on generative adversarial network screening image blocks

A technology for histopathology and type classification, applied in biological neural network models, instruments, character and pattern recognition, etc., can solve problems such as affecting the effect of training, difficulty in training densely connected network parameters, and inability to screen normal areas of benign images. Improve classification accuracy and efficiency, save storage space, and eliminate redundancy

Pending Publication Date: 2020-12-18
BEIJING UNION UNIVERSITY
View PDF9 Cites 7 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Its disadvantage is that it cannot screen the normal areas in benign images and malignant images, which affects the effect of training; and the densely connected network will have too many parameters and is not easy to train

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Breast cancer histopathological type classification method based on generative adversarial network screening image blocks
  • Breast cancer histopathological type classification method based on generative adversarial network screening image blocks
  • Breast cancer histopathological type classification method based on generative adversarial network screening image blocks

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0054] Such as figure 1 As shown, step 100 is executed to obtain an image data set of histopathological types of breast cancer. Step 110 is executed to perform preprocessing on the images of breast cancer histopathological types. This step includes the following sub-steps:

[0055] Step 11: Perform dyeing standardization processing on the breast cancer histopathological type image data set;

[0056] Step 12: dividing the breast cancer histopathological type image data set into a training set, a verification set and a test set; performing a random image block sampling strategy on the normal images in the training set and the verification set to obtain normal image blocks;

[0057] Step 13: Use the normal image block to train an unsupervised generation confrontation network, the generation confrontation network is composed of two confrontation modules, a generation network G and a discriminant network D, and the objective function of the generation confrontation network during...

Embodiment 2

[0080]For high-resolution breast cancer histopathological images, existing traditional machine learning methods and deep neural network models for direct analysis of whole-section digital histopathological images will lead to very complex architectural problems. In the past few years, Image patch-based methods for breast cancer histopathology image classification have achieved promising results in breast cancer histopathology datasets. However, it is very challenging to adopt an image patch-based method for breast cancer histopathological image classification, because there are benign and normal regions in malignant whole-section digital histopathological images, and there are normal regions in benign whole-section digital histopathological images, so only Partially extracted image patches are correctly labeled. In order to solve this problem of mislabeled image patches and further improve the classification accuracy. We propose a breast cancer histopathological type classifi...

Embodiment 3

[0099] In the research on the existing technology, it is found that there are the following shortcomings: it is impossible to screen the normal regions in benign images and malignant images, which affects the effect of training; and the densely connected network has too many parameters and is difficult to train.

[0100] In view of the above-mentioned shortcomings, the present invention increases the screening of normal areas in benign and malignant image blocks in the screening image block strategy to ensure more accurate data, and adopts a convolutional neural network based on a more compact network structure based on cyclic dense connections. The modification can solve the above problems, so as to achieve more accurate and efficient classification results of breast cancer histopathological images.

[0101] The present invention is mainly made of four modules:

[0102] (1) Breast cancer histopathological image preprocessing

[0103] (2) Based on the generated confrontation ...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention provides a breast cancer histopathological type classification method based on generative adversarial network screening image blocks, which comprises the following steps: acquiring a breast cancer histopathological type image data set, and further comprises the following steps: preprocessing breast cancer histopathological type images; enabling the generative adversarial network to screen normal regions in the benign image blocks and the malignant image blocks; enabling the generative adversarial network to screen benign regions in the malignant image blocks; and classifying thebreast cancer histopathological images by using a convolutional neural network based on cyclic dense connection. According to the invention, the improved unsupervised generative adversarial network isadopted to learn the data distribution of the normal pathology image and the benign tumor pathology image respectively, so that the benign tumor area and the normal area in the malignant tumor pathology image and the normal area in the benign tumor pathology image can be screened; and the possibility is provided for assisting doctors to diagnose the illness state more accurately and more quicklyto the maximum extent.

Description

technical field [0001] The invention relates to the technical field of image feature description, in particular to a method for classifying breast cancer histopathological types based on generating adversarial network screening image blocks. Background technique [0002] Breast cancer is the most common cancer among women worldwide, affecting approximately 2.1 million women each year. Breast cancer is a serious disease in which cancer cells grow unchecked in the body, beyond their cellular boundaries to invade adjacent sites or spread to other organs. In the United States, an estimated 276,480 new cases of invasive breast cancer and 48,530 new cases of noninvasive breast cancer will be diagnosed in women in 2020, according to the most recent data provided by the American Cancer Society. An estimated 42,170 women in the United States are expected to die from breast cancer by the end of 2020. [0003] Due to the high mortality rate of breast cancer, women are advised to unde...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06K9/62G06N3/04
CPCG06V2201/03G06N3/047G06N3/045G06F18/214G06F18/2415
Inventor 杨萍满芮季程雨芦博李欣桐
Owner BEIJING UNION UNIVERSITY
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products