Learning Dynamical Shape Prior for Level Set based Cell Tracking

Automated cell tracking in populations is very crucial for studying dynamic cell cycle behaviors. However, a high accuracy of each step is essential to avoid error propagation. In this paper, we propose an integrated three-component system to tackle this problem. We first model the temporal dynamics of shape change using an autoregressive model, which is used for estimating the shape and the location of the current object. We then segment the cell using an active contour model starting from the predicted shape. Finally, we identify its phase using a Markov model. The phase information is also used for finetuning the segmentation result. We applied this approach for tracking HeLa H2B-GFP cells and tested it in different aspects. Highly accurate validation results confirm the usefulness of our integrating approach and show its robustness and the essentiality of each component.