Feature-aided multiple-hypothesis tracking and classification of biological cells

Our work is motivated by emerging technology for dielectrophoresis-based amplification of cell kinematic responses, allowing for cell classification based on motion in addition to static cell features. Accordingly, the approach requires high-performance multi-target tracking. This paper discusses our approach to the problem, which involves suitable enhancements to track-oriented multiple-hypothesis tracking algorithms. Additionally, we apply a generalized likelihood ratio test for cell classification. Preliminary simulation-based testing and laboratory experimentation with cell and bacteria cultures show encouraging results.