A novel method for automatic single molecule tracking of blinking molecules at low intensities

Single molecule tracking provides unprecedented insights into diffusional processes of systems in life and material sciences. Determination of molecule positions with high accuracy and correct connection of the determined positions to tracks is a challenging task with, so far, no universal solution for single fluorescing molecules tackling the challenge of low signal-to-noise ratios, frequent blinking and photo bleaching. Thus, the development of novel algorithms for automatic single molecule fluorescence tracking is essential to analyse the huge amount of diffusional data obtained with single molecule widefield fluorescence microscopy. Here, we present a novel tracking model using a top-down polyhedral approach which can be implemented effectively using standard linear programming solvers. The results of our tracking approach are compared to the ground truth of simulated data with different diffusion coefficients, signal-to-noise ratios and particle densities. We also determine the dependency of blinking on the analysed distribution of diffusion coefficients. To confirm the functionality of our tracking method, the results of automatic tracking and manual tracking by a human expert are compared and discussed.