Cell sorting method for affine propagation clustering

Abstract

The invention relates to a cell sorting method for an affine propagation clustering, which comprises the following steps: (1) selecting a circularity parameter C and a rectangularity parameter R of a cell image, designing a sample coordinate X sample=lambda C+(1-lambda) R, selecting an area parameter Area of the cell image as another sample coordinate Y sample and selecting a nuclear-cytoplasmic ratio parameter prop of the cell image as another sample coordinate Z sample, wherein lambda represents a prior input value; (2) using euclidean distance of the three-dimensional sample coordinates as sample distance, wherein a diagonal value of an S matrix of the affine propagation clustering is an average value of distances among the samples; (3) under the initial condition, setting an membership grade matrix A(i, k)=0, updating a matrix R and updating a matrix A; and (4) stopping after the number of iteration times is set and obtaining different types of cells from a sorting result. The invention provides the cell sorting method for the affine propagation clustering, which is suitable to process mass data, has excellent real-time property and can effectively carry out cell sorting.

Description

Technical field [0001] The invention relates to image processing, biomedicine, computer vision, and calculation methods, in particular to a cell classification method after cell image segmentation. Background technique [0002] The purpose of cell image segmentation is to extract the cell body. After the cell body is successfully segmented, the image information can be converted into a numerical value, which is the work completed in Chapter 4 of this article. Although the segmentation effect of different segmentation methods is quite different, they can obtain cell monomers, and then complete the conversion of cell monomers into digital statistics, which are morphological parameters. Most research results only put forward the idea of ​​segmentation or statistical morphological parameters, which causes researchers to face massive amounts of data, which increases the difficulty of research and analysis time. [0003] Commercial cell morphology analysis software-IMT tissue morphology...

AI Technical Summary

Problems solved by technology

[0005] In order to overcome the shortcomings of the existing cell morphology analysis methods, such as high computational complexity, poor real-time performance, and inability to classify cells, the present invention provides an affine propagation aggregation method suitable for processing massive data, good real-time performance, and capable of effectively classifying cells. class of cell sorting methods

Images