Image processing for super-resolution localization in fluorescence microscopy

Localization of a single fluorescent particle with sub-diffraction limit accuracy is a key merit in fluorescence microscopy. Implementation of nonlinear filtering algorithms prior the localization process can improve the localization accuracy of standard existing methods and also enable the localization of overlapping particles, allowing the use of increased fluorophore activation density, and thereby increased data collection speed. In this paper we present the use of an image decomposition algorithm termed K-factor which reduces an image into a nonlinear set of contrast-ordered decompositions whose joint product reassembles the original image. The K-factor technique is implemented on the images, prior to the localization of the fluorescent probes. Numerical simulations of fluorescence data with random probe positions, and especially at high densities of activated fluorophores demonstrated an improvement in the localization precision with compare to single fitting techniques. Implanting the proposed concept also on experimental data of cellular structures yielded the theoretically predicted resolution enhancement.