Study on Cell Cluster Splitting

In the filed of biomedical engineering, it is often to use image technique for cell analysis. To recognize cells, one basic and hardest task is for cell delineation, and the cell clusters must be decomposed. In this paper, a novel algorithm based on shape information is proposed for splitting cell clusters. In the algorithm, it integrates morphological smoothing with holes filling in a pre-processing stage. Then the cell clusters are identified through a polygon approximation. Finally the cluster decompose is implemented, which consists of the detecting concave points on the contours and determining the decomposing lines. The algorithm can not only split the simple cell clusters, but also complicated clusters. In addition, this algorithm can be adopted for other applications, where separation between touching and overlapping particles is required.