Tracking of non-dividing cells by using generalized Voronoi diagram

Cell tracking is an important technique to study cell migration. However, to obtain good tracking results is still a challenging task. False positives and false negatives are two main factors that affect cell tracking accuracy. It is desirable to have a good method to eliminate false positives and false negatives effectively. We propose to use generalized Voronoi diagrams (GVDs) to perform tracking for non-dividing cells in a key point evolving based algorithm. This method can realize tracking association for multiple objects based cell racking and eliminate false positives and false negatives effectively based on a defined cost function. We investigated under what conditions our algorithm would work, and what were the results when the assumptions for the algorithm were violated. This provides the relationship between the cell migration speed and movie temporal resolution. We further validated that our algorithm works well under reasonable assumptions. We tested our algorithm with both 2D+time and 3D+time data sets. In conclusion, the use of Generalized Voronoi Diagram in cell tracking is an effective technique to increase tracking accuracy. The temporal resolution when acquiring movies should be ensured in order to achieve good tracking results.