Figure 3: Data organization and formats.

Videos, metadata and the position over time of tracked cells are organized as the conceptual Entity-Relationship model described in (a), corresponding to the logical database (optimized version) depicted in (b). A video belongs to a Collection, depicts one or more Problematic and includes an Image series. The Image entity is double-framed because it is a weak entity, which depends on the Video entity. A Cell has one type and one unique identifier. One Video tracks one (or more) Cell, every Cell being depicted by the Track association at a given timestamp (t) and in a spatial position (x, y, z) of that Video. The VisibleIn association further describes the channel of the video in which a cell is visible. The logical database is derived from the conceptual model and then optimized for read-access. The cells_positions table stores the instantaneous coordinates of each cell and is pre-exported in one or more CSV text files for each video. Imaging data are logically stored as TIFF image series with a specific filename.