FIGURE 2.

Computational neural network analysis pipeline. (A) Overview of the sequential CNN multi-object tracking pipeline Oligo-Track (top). CNNs marked in green. Overview of heuristic baseline method (orange) for comparison to Oligo-Track (bottom). Compatibility with optional syGlass curation provides validation of tracking in both pipelines (blue). Curated tracks can also be reintroduced into training pipeline for refinement of CNNs. (B) Seg-CNN pre-processing extracts cropped regions from larger volumes with 50% overlap for computational efficiency at each timepoint t to t + n. Cropped regions are restitched to form timeseries. Scale bar: 50 μm.