Deep attention networks for automated collective behavior discovery in epithelia

This dataset contains the raw timelapse movie data and parsed cell trajectory data, in the form of XML files, for the experimental data as described in our paper, "Deep attention networks for automated collective behavior discovery in epithelia". Timelapse movies (both transmitted light images and fluorescence images highlighting cell nuclei) are provided for human umbilical vein endothelial cells (HUVEC), Madin-Darby Canine Kidney (MDCK) cells, and MDA-MB-231 cells (a breast cancer line). XML files containing cell trajectory data (2D points in time) are provided for these same cell types. Trajectories are taken from the bulk region of the tissue (see paper) and XMLs are generated using the TrackMate plugin in ImageJ/FIJI. For MDCKs, additional XMLs are provided for cell trajectories taken from the edge region (see paper).