Graph Neural Network for Cell Tracking in Microscopy Videos

We present a novel graph neural network (GNN) approach for cell tracking in high-throughput microscopy videos. By modeling the entire time-lapse sequence as direct where instances are represented by its nodes and their associations edges, we extract set of trajectories looking maximal paths graph. This is accomplished several key contributions incorporated into an end-to-end deep learning framework. exploit metric algorithm to feature vectors that distinguish between different biological cells assemble same instances. introduce new GNN block type which enables mutual update node edge vectors, thus facilitating underlying message passing process. The concept, whose extent determined number blocks, fundamental importance it ‘flow’ information edges much behind neighbors consecutive frames. Finally, solve classification problem use identified active construct cells’ tracks lineage trees. demonstrate strengths proposed applying 2D 3D datasets types, imaging setups, experimental conditions. show our framework outperforms current state-of-the-art methods on most evaluated datasets. code available at https://github.com/talbenha/cell-tracker-gnn .