Motion Sensing Superpixels (MOSES): A systematic framework to quantify and discover cellular motion phenotypes

Cellular motion is fundamental in tissue development and homeostasis. There is strong interest in identifying factors that affect the interactions of cells in disease but analytical tools for robust and sensitive quantification in varying experimental conditions for large extended timelapse acquisitions is limited. We present Motion Sensing Superpixels (MOSES), a method to systematically capture diverse features of cellular dynamics. We quantify dynamic interactions between epithelial cell sheets using cell lines of the squamous and columnar epithelia in human normal esophagus, Barrett9s esophagus and esophageal adenocarcinoma and find unique boundary formation between squamous and columnar cells. MOSES also measured subtle changes in the boundary formation caused by external stimuli. The same conclusions of the 190 videos were arrived at unbiasedly with little prior knowledge using a visual motion map generated from unique MOSES motion "signatures". MOSES is a versatile framework to measure, characterise and phenotype cellular interactions for high-content screens.