Unlock instant, AI-driven research and patent intelligence for your innovation.

A method for detecting cell mitosis in breast cancer scoring system

A technology of mitosis and detection method, which can be applied in the fields of analyzing materials, image analysis, measuring devices, etc., can solve problems such as indistinguishability, and achieve the effect of solving the shortage of training samples

Inactive Publication Date: 2020-04-24
成都知识视觉科技有限公司
View PDF6 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

At most stages, mitotic and non-mitotic nuclei are very similar, or are similar to other dark blue dots, and it is difficult for untrained people to distinguish them
Another complication is that at the end of mitosis the nucleus splits into two dark blue blobs, i.e. one mitosis will have two distinct nuclei but they are not yet a single cell in the full sense

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

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0026] Embodiment 1: a method for detecting cell mitosis in a breast cancer scoring system, comprising the steps of:

[0027] A. The pathologist manually marks the ROI area in the digital slice of breast cancer to obtain the image of cell mitosis;

[0028] B. Read in the image of the digital slice file, cut the image into small pieces, and obtain the information of whether the small piece of image contains mitosis by querying the information in the file marked by the pathologist, and mark the position of mitosis, thus obtaining two Class sample set, namely: mitotic samples, not including mitotic samples;

[0029] C. Design cell mitosis generation model;

[0030] D. Using the samples without mitosis as the simulation training set, the mitotic cell samples are simulated;

[0031] E. Use the simulated mitosis samples as training samples, and the mitosis samples marked by doctors as verification samples, start the CNN neural network, start training, let the network automatically...

Embodiment 2

[0032] Embodiment 2: a method for detecting cell mitosis in a breast cancer scoring system, comprising the steps of:

[0033] A. The pathologist manually marks the ROI area in the digital slice of breast cancer to obtain the image of cell mitosis;

[0034] B. Read in the image of the digital slice file, cut the image into small pieces, and obtain the information of whether the small piece of image contains mitosis by querying the information in the file marked by the pathologist, and mark the position of mitosis, thus obtaining two Class sample set, namely: mitotic samples, not including mitotic samples;

[0035] C. Design cell mitosis generation model, cell expansion coefficient p1, corrosion depth p2, ellipse corrosion long-short axis ratio p3, corrosion angle p4, use mGA to generate multiple sets of [p1, p2, p3, p4] values, p1, p2, Both p3 and p4 are vectors of a certain length n;

[0036]D. Take the samples without mitosis as the simulation training set, and make the fol...

Embodiment 3

[0038] Embodiment 3: a method for detecting cell mitosis in a breast cancer scoring system, comprising the steps of:

[0039] 1) The pathologist selects the ROI region (Region of Interest) in the breast cancer digital slice (Whole Slide Image, WSI) for manual labeling to obtain images of cell mitosis;

[0040] 2) Cut the image into small patches (patches), for example: the size of each patch is 128x128 pixels, by querying the information in the pathologist’s annotation file, get the information of whether the patch contains mitosis, and mark the position of mitosis , thus obtaining two types of sample sets, namely: mitotic samples and non-mitotic samples;

[0041] 3) Design cell mitosis generation model, cell expansion coefficient p1, corrosion depth p2, ellipse corrosion long-short axis ratio p3, corrosion angle p4, in order to simulate multiple possibilities, p1, p2, p3, p4 are all vectors of a certain length n (such as n=1000);

[0042] 4) Use mGA to generate multiple set...

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 discloses a cell mitosis detection method of a breast cancer scoring system. The method comprises that a pathologist marks an ROI in a digital bread cancer slice manually, and a cell mitosis image is obtained; a digital slice file image is read in, information in a file marked by the pathologist is inquired to obtain whether an image block includes information of mitosis, a position of mitosis is marked, two types of sample sets are obtained, and a cell mitosis generating model is designed; samples not including mitosis for a simulated training set, and mitosis cell samples are obtained by simulation; and the mitosis samples obtained by simulation serve as a training sample set, the mitosis sample marked by the pathologist service as a verification sample, CNN is started, training is started, features of mitosis is learned automatically by the network, and an identification model is established. The method of the invention can be used to solve the problems that training samples are insufficient in amount and a routine image identification method is not high in versatility or robustness.

Description

technical field [0001] The invention relates to a method for detecting cell mitosis in a breast cancer scoring system. Background technique [0002] The female mammary gland is composed of skin, fibrous tissue, mammary glands and fat. Breast cancer is a malignant tumor that occurs in the mammary gland epithelial tissue. 99% of breast cancer occurs in women, and only 1% in men. Breast cancer is the number one common malignant tumor in women. [0003] The global incidence of breast cancer has been on the rise since the late 1970s. 1 in 8 women in the U.S. will develop breast cancer in her lifetime. China is not a country with a high incidence of breast cancer, but it should not be optimistic. In recent years, the growth rate of the incidence of breast cancer in my country is 1 to 2 percentage points higher than that of countries with a high incidence. [0004] During the diagnosis of breast cancer, pathologists first extract tissue samples from the lesion area of ​​the brea...

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
Patent Type & Authority Patents(China)
IPC IPC(8): G06T7/00G01N15/10
CPCG01N15/10G01N2015/1006G06T7/0012G06T2207/10056G06T2207/20021G06T2207/20081G06T2207/20084G06T2207/20104G06T2207/30068G06T2207/30096
Inventor 蒲洋向飞
Owner 成都知识视觉科技有限公司