Robust Cell Nuclei Tracking Using Gaussian Mixture Shape Model

Abstract

The life cell microscopic imaging is a standard approach for studying of cancer cell morphology and behaviour during some treatment. In the dense cell cultures, tracking each cell nucleus is challenging task due to cell overlap and interactions. Moreover, for time-lapse sequences (lasting typically 20-30 hours) the robust automatic cell tracking is needed. This paper describes new method for fluorescence nuclei tracking based on Gaussian mixture model (GMM), and additionally, GMM modification allowing application to the images is also introduced. Method is mainly designed for robustness - tracking the highest possible number of nuclei in the whole sequence. Proposed algorithm proved to by very reliable with 80% of correctly tracked nuclei.