Fast, Simultaneous Multiple Fluorophore Fitting in Single Molecule Super-Resolution Imaging

Abstract

Single molecule localization based super-resolution (SML-SR) imaging techniques require repeated localization of many single fluorophores. If, during analysis, successful localization does not require isolated fluorophores, the performance can be improved in one or more of several metrics that result in higher single-frame density of active fluorophores. For example: Data acquisition time can be reduced; A larger number of fluorophores can be localized; There is a higher tolerance on labeling density; and Dyes with higher minimum duty cycle can be used.We have developed a method that uses the maximum likelihood estimator to localize multiple fluorophores simultaneously within a fitting sub region. We find that for a fitting region of size approximately 6 sigmaPSF (where sigmaPSF parameterizes a 2D Gaussian PSF model) localization of up to 5 fluorophores provides a good compromise between fit accuracy and analysis time. For speed, the algorithm is implemented on Graphics Processing Unit (GPU) architecture in a manner similar to our previous single molecule analysis (Smith, Nat. Methods 7, 373-375 2010) and achieves near real-time analysis speed.We show the performance of multiple fluorophore fitting as a function of (1) maximum allowed number of fitted fluorophores and (2) single-frame active emitter density. We describe the details of the algorithm that allow robust fitting, the details of the GPU implementation, and the other imaging processing steps required for the analysis of (SML-SR) data sets. As a demonstration, we show that our new multi-fluorophore super resolution imaging method reveals actin structure in a HeLa cell under conditions where a high single-frame fluorophore density results in poor reconstructed images using conventional single fluorophore based super resolution imaging techniques.