Pushing the performance of image scanning microscopy to its limits with maximum likelihood reconstruction

Abstract

Fast and sensitive detector arrays make Image Scanning Microscopy (ISM) the natural successor of confocal microscopy. Indeed, ISM enables super-resolution at an excellent signal-to-noise ratio. Optimizing photon collection requires large detectors and so more out-of-focus light is collected. Nonetheless, the ISM dataset inherently contains information on the axial position of the fluorescence emitters. We exploit such information to directly invert the cmresponding physical model with a maximum-likelihood approach and reassign the signal in the thr ee dimensions, improving the signal-to-background ratio and resolution. We validated our method on synthetic and experimental images; these latter acquired with a custom setup equipped with a single photon avalanche diode array detector. Moreover, our method is compatible with recent developments in ISM data processing and requires minimal knowledge of physical parameters.