Multiple neutrophils tracking in vitro array using high-order temporal information

It is crucial to dynamically analyze the movement of neutrophils (a type of white cell) during the process of inflammation. However, manually tracking and analyzing the cells is a time-consuming task due to the large volume of cells and similar appearance. To facilitate neutrophils analysis and address the issues mentioned above, we propose to leverage high-order temporal information as a cue to track neutrophils. A tensor-based approach is introduced to encode the high-order motion pattern and appearance variation for multi-frame multicell association. To evaluate the proposed method, we collected 354 sequences of cells from 200 frames of microscopic images. We conduct a systematic study on the collected data and show significant performance improvement over other solutions.