Research On Stem Cell Segmentation And Tracking Method

Stem cell is a kind of pluripotent cells with self-renewing ability.Under certain conditions,they have the ability to differentiate into other cells and tissues.Therefore,they have great application prospects in the fields of regenerative medicine and repairing tissue damage.With the growth of cell image data,the traditional manual analysis method is more and more difficult to solve the problem.Digital image processing technology,as a new automatic processing method,can not only solve the impact of subjective analysis,but also save a lot of human resources.Therefore,the use of image processing technology to analyze cell images has become an important research direction.This paper mainly studies the stem cell images based on phase contrast microscopy.The main contents include: phase contrast microscope image segmentation,cell tracking and cell analysis integrated software system.Specifically divided into the following three aspects:1.Aiming at problem of halo around the cell imaging of phase contrast microscope and the difficulty of traditional methods to segment,this paper study two methods which can segment the image of phase contrast microscope and eliminate the halo: based on the difference between imaging principle of image restoration,and segmentation method based on the halo correction.Firstly,the two methods are compared.The latter method has higher processing speed and accuracy than the former method when processing cell images with low confluence degree,but the latter depends on the initial coarse segmentation threshold value.Secondly,the distribution law of phase difference microscope image is analyzed,and an adaptive threshold segmentation method combined with halo correction is proposed,which can calculate the optimal segmentation threshold for different images adaptively,and then accurately segment cells by combining light and shadow correction,so as to improve the segmentation accuracy.2.This paper study a feature matching stem cell tracking algorithm based on mitotic detection,analyzes the morphological changes during cell mitosis,and its features to realize the detection of cell mitosis by convolutional neural network.At the same time,the principle of multi-target matching tracking is analyzed,and mitosis is introduced into the multi-target matching tracking method,which can significantly reduce the computational cost of the method without reducing the accuracy of cell tracking in the early stage.3.An integrated software system for cell analysis is designed to realize human-computer interaction and make cell analysis more convenient.This paper is based on Matlab GUI graphics program design the main functional modules include: segmentation module,tracking module.Segmentation module mainly calculates the static information of cells,such as cell location and area,while the tracking module mainly calculates the dynamic information of cells,such as cell speed and acceleration.The software system can help researchers better analyze the behavior of cells.