Medical Image Analysis Based On Multi-target Tracking

Medical imaging has been an important direction of medical research field. Scholars are dedicated to medical imaging research studies over the years. The recognition and detection of medical images have been a great concern. Many traditional work of medical image recognition is based on artificial approach, also it is time-consuming and the accuracy is largely dependent on the expertise of medical personnel. As the rapid development of computer technology lately, we hope to achieve fast and efficient computer assisted medical image recognition. Target tracking technology has been a key processing technology of moving target, which can be referred to medical slice images. After having the trajectory between frames, and then combining the trajectory and the characteristics, we do analysis of every target to the medical image recognition. This paper is about medical image analysis based on multi-target tracking.It is difficult to use a single tracking method to track all the cells of the sequence images due to the cell movement rules can’t be found. The paper improves the movement model under the Kalman framework. We use "current model" which can adjust adaptively to the movements of cells based on the characteristics of each cell, and especially for the cells that has large range of motion between frames. We generally ignored the issue of splitting consideration and only to consider the results of the whole sequence of images before. In the paper, we add features based on size and location of the cells to detect the splitting cells. It can determine the splitting cells in the images that are not very complex.The traditional manual method of identifying lymph nodes cannot meet a large number of slices for image processing. So after the analysis of the stomach images, we give a method to detect the lymph nodes based on objective tracking method. We use the tracking method to get the trajectory of each target in the images, which use both the trajectory and the characteristics of the targets.This research is supported by the National Natural Science Foundation of China (No.61173092), the Fundamental Research Funds for the Central Universities (NO. JY10000902045), higher school subject innovation engineering plan (NO. B07048).