Effective models for segmentation and shape tracking of cells

Abstract

Several techniques were developed to give high efficiency for segmentation and shape tracking of cells. Owing to the large amount of data and complexity, the techniques were not suitable for analyzing both 2D and 3D data. The proposed technique deals with a fast and robust method that can be applied to time-lapse input series. In this case, the algorithm consists of four steps for the analysis of each frame in the time-lapse input series. The input frame is segmented in the first step; the cell boundaries are detected in the second step; the total cell area is determined in the third step, and then finally the cell area of each frame is compared with the results of successive frames. The robust and reliable steps in the algorithm clearly describe the variation in shapes of cells. Extensive simulation results show that the accuracy of the proposed algorithm is high and it substantially outperforms the other existing algorithms.