A new active contour model and its application on cell segmentation

This paper presents a novel active contour model. In our model, the contour points are described in polar coordinates and limited to move along their radial directions. When implemented with greedy algorithm, the proposed model has lower computation cost than the traditional snake described in Cartesian coordinates. To broaden the attraction range of the object boundary, improvements have been made on the internal energy and the external energy based on the polarized description of our model. The results of applying our model to segment cell nucleus of biopsy cell images show that it can track cell nucleus boundary efficiently and our model has superior performance than the traditional snake.