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

Using Masks to Improve Classification Performance of Convolutional Neural Networks for Cancer Cell Screening Applications

A technology of convolutional neural network and cell classification, which is applied in the field of improving the performance of convolutional neural network in cell classification, and can solve problems such as incorrect learning and misclassification of CNN

Active Publication Date: 2021-07-06
HONG KONG APPLIED SCI & TECH RES INST
View PDF9 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

These irrelevant objects may cause CNN to learn incorrect features, resulting in misclassification

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
  • Using Masks to Improve Classification Performance of Convolutional Neural Networks for Cancer Cell Screening Applications
  • Using Masks to Improve Classification Performance of Convolutional Neural Networks for Cancer Cell Screening Applications
  • Using Masks to Improve Classification Performance of Convolutional Neural Networks for Cancer Cell Screening Applications

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0043] As used herein, a training image means an image used to train a CNN, and a test image means an image processed by a CNN or to be processed for classification. Furthermore, herein in the specification and appended claims, it is to be understood that "an image comprising cells" means that the image comprises sub-images of cells rather than that the image comprises solid cells.

[0044] The present invention is about classifying cells by using CNN. Important applications of this classification include the screening of cancer cells and the screening of precancerous abnormalities. However, the present invention is not limited to applications for cancer cell screening and precancerous abnormality screening only. The present invention finds use in other medical and biological applications. Furthermore, cells referred to in this classification are not limited to be of human origin only. Cells can be of animal (eg equine) or plant origin. In the following, the invention is e...

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

In cancer screening, a patient's cells are classified by a convolutional neural network (CNN) to identify abnormal cells. In one approach, a mask having a center that is more transparent than the perimeter of the mask is used to mask an input image containing cells of interest to produce a masked image. Because cells are usually located near the center of the image, and because images often contain extraneous objects near the periphery of the image, such as normal cells and microbes, by using masked images instead of original images, the loss of extraneous objects is reduced when training CNNs and when classifying caused interference. In another approach, masking is applied to the feature maps prior to classification. In CNNs, this masking is accomplished by convolving each feature map with a convolution kernel to produce an intermediate feature map, and then truncating its surrounding regions to produce a downsized feature map.

Description

[0001] list of abbreviations [0002] ADC cervical adenocarcinoma [0003] AGC atypical glandular cell abnormalities [0004] AIS adenocarcinoma in situ [0005] ASC-H Atypical squamous cells - does not rule out HSIL [0006] ASC-US atypical squamous cell abnormality of undetermined significance [0007] CNN convolutional neural network [0008] HSIL high-grade squamous intraepithelial lesion [0009] LSIL low-grade squamous intraepithelial lesion [0010] SCC squamous cell carcinoma [0011] TBS Bethesda System [0012] WSI Full View Digital Slice technical field [0013] The present invention relates to methods for improving the performance of convolutional neural networks (CNNs) in cell classification. More particularly, the present invention relates to methods for improving the classification performance of CNNs used in cancer cell screening. Background technique [0014] Cervical cancer is cancer that arises in a woman's cervix. The routine method of cervical...

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): G06K9/62G06N3/04
CPCG06N3/045G06F18/241
Inventor 何学俭王陆
Owner HONG KONG APPLIED SCI & TECH RES INST