Deep Particle Tracker: Automatic Tracking of Particles in Fluorescence Microscopy Images Using Deep Learning

Tracking of particles in fluorescence microscopy image sequences is essential for studying the dynamics of subcellular structures and virus structures. We introduce a novel particle tracking approach using an LSTM-based neural network. Our approach determines assignment probabilities jointly across multiple detections by exploiting both short and long-term temporal dependencies of individual object dynamics. Manually labeled data is not required. We evaluated the performance of our approach using image data of the ISBI Particle Tracking Challenge as well as real fluorescence microscopy image sequences of virus structures. It turned out that the proposed approach outperforms previous methods.