Automated Neuron Tracking in C. elegans Using Segmentation and Graph Neural Networks

Tracking neuronal activity in C. elegans is critical for understanding neural dynamics, but challenges arise due to brain deformation and overlapping neurons. Current methods struggle with accurate neuron localization and tracking over time. We propose a novel approach integrating segmentation and graph neural networks (GNNs) to enhance neuron tracking. By using segmentation to extract neuronal features and establishing anchor points, our method improves image registration and neuron matching. A GNN model is employed to match neurons across frames, significantly improving tracking precision in deformable C. elegans brains. This system offers a promising advancement in neuron tracking accuracy.