Dynamic sub-cellular behavior study in high content imaging using a multi-threshold approach

Clathrin-coated pits and vesicles are found in all nucleated cells, from yeast to humans. They represent an important means by which proteins and lipids are removed from the plasma membrane and transported to an internal compartment. The biolgolical question is that [5]: is there "hot spots" for the formation of clathrin-coated pits, or do pits and arrays form randomly on the plasma membrane? To answer this question we would like to track many hundreds of individual pits as they form, and track whether the pit is successful in forming a vesicle. In this paper, we propose a novel polynomial fitting based multi-threshold approach for particles imaging segmentation. Simulated Annealing is employed to find the optimal thresholds and Auto-regression model is used to reduce the search space. A combined aspect ratio and size of particles are proposed to isolate touching spots. In the particle tracking, a combined centroid and shape based method is proposed to deal with the ambiguous association problem. The proposed approach has been shown to identify and track lots of particles successfully.