Deep Neural Network for Combined Particle Tracking and Colocalization Analysis in Two-Channel Microscopy Images

Analyzing protein dynamics in multi-channel fluorescence microscopy data is important to understand biological processes. We present a novel deep learning approach for combined particle tracking and colocalization analysis in two-channel microscopy image sequences. The approach is based on a convolutional long short-term memory network and exploits colocalization information to improve tracking. Short and long-term temporal dependencies of object motion as well as image intensities are taken into account to compute assignment probabilities jointly across multiple detections. Colocalization probabilities are also determined by the neural network. We evaluated the performance of the proposed approach based on synthetic images and real two-channel fluorescence microscopy data. It turned out that our approach outperforms previous methods.