Cell tracking is one of the most challenging time-lapse image-analysis tasks. It is the basis for analyzing cells at the single-cell level and study cell motility in various contexts, for example cancer cell invasion, immune cell migration, and embryo development.

‍

‍

The solution provided here can be used to track cells, analyze cell division, and quantify various features of the cells over time (e.g., size, sphericity, diameter). First, a deep-learning model was trained to segment each cell separately (using instance segmentation). Then the model was used to automatically segment all time points imaged in a multi-well plate format and then all cells were tracked. Furthermore, cells were classified into two groups based on size or marker expression.

‍