ShapeSpaceExplorer: Analysis of morphological transitions in migrating cells using similarity-based shape space mapping

Abstract

Here we describe the development of ShapeSpaceExplorer, an interactive software for the extraction, visualisation and analysis of complex 2D shape series. We also demonstrate its application to the analysis of cell morphology changes during cell migration. Cell migration is essential for many physiological and pathological processes. Intracellular force generation and transmission of these forces to extracellular structures or neighbouring cells drives cell migration. The emergent property of the processes driving cell migration is a change in cell shape. We describe a machine learning approach to understand the relationship of cell shape dynamics and cell migration behaviour. Our algorithm analyses cell shape from time-lapse images and learns the intrinsic low-dimensional structure of cell shape space. We use the resultant shape space map to visualise differences in cell shape distribution following perturbation experiments and to analyse the quantitative relationships between shape and migration behaviour. The core of our algorithm is a new, rapid, and landmark-free shape difference measure that allows unbiased analysis of the widely varying morphologies exhibited by migrating mesenchymal cells. We used our method to predict cell turning from dynamic cell shape information. ShapeSpaceExplorer can be applied widely to visualise and analyse cell morphology changes during development, the cell cycle and stress response, but also to the outlines of clusters, tissues and inanimate objects.