Keele Research Repository

This paper proposes the combined use of advanced texture based analytics and the Normalised Cross Correlation procedure to facilitate in vitro cells tracking with high-throughput time-lapsed video microscopy. The effectiveness of this technique was examined in the context of unlabelled phase contrast (PC) imaging, with the goal to investigate the life cycle of back disc cells. Computationally, the typically low contrast PC image adds to the challenges of cell tracking, which is often exacerbated by the highly varying cell morphology. To accurately determine the individual cell locations, the proposed algorithm computes the dynamic set of textural features for each cell, allowing its accurate identification in successive image/time frames. More importantly, such features also help calculate the location trajectory of every cell, for which further quantitative analysis can be developed, as proof of principle, to facilitate granulometric studies of vesicles present in these individually tracked cells. Preliminary results were compared to the manually constructed ground truth, allowing benchmarking of the proposed tracking algorithm in parallel with a freely available cell tracking platform.