Multiple Hypothesis Tracking for Cluttered Biological Image Sequences

International audienceIn this paper, we present a method for simultaneously tracking thousands of targets in biological image sequences, whichis of major importance in modern biology. The complexity and inherent randomness of the problem lead us to propose a unifiedprobabilistic framework for tracking biological particles in microscope images. The framework includes realistic models of particlemotion and existence, and of fluorescence image features. For the track extraction process per se, the very cluttered conditionsmotivate the adoption of a multiframe approach which enforces tracking decision robustness to poor imaging conditions and to randomtarget movements. We tackle the large-scale nature of the problem by adapting the Multiple Hypothesis Tracking algorithm to theproposed framework, resulting in a method with a favorable trade-off between the model complexity and the computational cost of thetracking procedure. When compared to the state-of-the-art tracking techniques for bioimaging, the proposed algorithm is shown to bethe only method providing high quality results despite the critically poor imaging conditions and the dense target presence. We thusdemonstrate the benefits of advanced Bayesian tracking techniques for the accurate computational modeling of dynamical biologicalprocesses, which is promising for further developments in this domain