Similarity analysis of cell movements in video microscopy

Modern 3D+T video microscopy techniques enable biologists to acquire data of living organisms with unprecedented resolution in time and space. These datasets contain a wealth of biologically relevant and quantifiable information, e.g. the movements of all individual cells in a complex organism. However, extraction, validation, and analysis of this information are both challenging and time-consuming. In this paper, we present a computational technique that classifies and validates similar patterns of cell movements and cell divisions in organisms that consist of up to thousands of cells. Our algorithm determines tracking paths of traced cells that exhibit similar features and shape structures. These similarity values are assigned to our cluster algorithm that clusters paths into groups of coherent behavior. The data can be interactively explored in 2D projections and a 3D cell movement representation. For the first time, this visualization allows biologists to exhaustively assess similarities and differences in division patterns and cell migration on the scale of an entire organism. For validation, we applied our method on a synthetic dataset and two real datasets including zebrafish periods from blastula stage to early epiboly and growing zebrafish tail. We show that our method succeeds in detecting similarities based on shape and cell-movement based features.