Capillary-driven microfluidic chips have emerged as promising platforms for point-of-care diagnostics, offering portable, inexpensive, and pump-free operation. Accurate tracking of cell flow in these systems is vital for quantitative applications such as on-chip cytometry, cell counting, and biomechanical analysis. However, tracking in capillary-driven devices is challenging due to rapid cell displacements, flow instabilities, and visually similar cells. Under these conditions, conventional tracking algorithms such as TrackPy, TrackMate, SORT, and DeepSORT exhibit frequent identity switches and trajectory fragmentation. Here, we introduce FloCyT, a robust, high-speed centroid tracking tool specifically designed for capillary-driven and microfluidic flow. FloCyT leverages microchannel geometry for tracking and uses anisotropic gating for association, global flow-aware track initialisation, and channel-specific association. This enables precise tracking even under challenging conditions of capillary-driven flow. FloCyT was evaluated on 12 simulated and 4 real patient datasets using standard multi-object tracking metrics, including IDF1 and MOTA, ID switches, and the percentage of mostly tracked objects. The results demonstrate that FloCyT outperforms both standard and flow-aware-modified versions of TrackPy and SORT, achieving higher accuracy, more complete trajectories, and fewer identity switches. By enabling accurate and automated cell tracking in capillary-driven microfluidic devices, FloCyT enhances the quantitative sensing capability of image-based microfluidic diagnostics, supporting novel, low-cost, and portable cytometry applications.
Maurya et al. (2025) studied this question.