Speeding up Collective Cell Migration Using Deep Reinforcement Learning

摘要

Collective cell migration is a significant and complex phenomenon since it influences many fundamental biological processes. The coordination between leader cell and follower cell impact the speed of collective cell migration. However, there are still very few papers to study the effect of the stimulus signal released by the leader on the follower. Using 3D time-lapse microscopy image to keep track of the process of cell movement provides an unprecedented opportunity to systematically investigate and analysis collective cell migration. Traditional approach to study the process always based on reality scene, but are too time-consuming as the number of cells grows exponentially. Agent-based modeling is a robust framework that approximates cells as isotropic, elastic, and adhesive objects. In this paper, we use the agent-based modeling framework to build a simulation platform for cell movement. Its goal is to construct a biomimetic environment to prove the importance of stimulus signals between leader cell and follower cell. We use the recent popular deep reinforcement learning to train cells and to control the quantity of signal. By experimenting on single-follower and multi-follower, we get the conclusion that the number of stimulation signals is proportional to the speed of collective cell movement. This type of research provides a more diverse approach and thinking to study biological issues.