Optic Flow Providing External Force for Active Contours in Visually Tracking Dense Cell Population

Abstract

Intense current research requires quantitative analysis of cell behaviours in dense cell populations. The low contrast cellular image quality, diversity of cell shapes, frequent cell interactions, and complex cell motions all pose significant problems to the efficient and robust cell tracking in phase contrast cellular images. We have proposed an automated cell tracking system based on active contours for tracking cell deformation and movement. The pyramidal optic flow scheme is exploited for providing external motion force to guide active contour evolution, and thus helps to address the particular difficulty in tracking relatively fast moving cells in dense cell population. We have evaluated the proposed framework on one real cellular dataset and proved an 80.2% tracking accuracy.