A novel fluorescence microscopy image deconvolution approach

Abstract

Superresolution optical fluctuation imaging (SOFI) is an attractive and affordable alternative to more established superresolution imaging methods. It provides moderate resolution enhancement and an efficient estimation of the optical point spread function (PSF). Moreover, further resolution enhancement could be achieved by deconvolution of the SOFI image. In this paper, we propose a novel image deconvolution approach based on the shearlet transform and the fractional-order total variation (FOTV) to further improve SOFI images. Since SOFI PSF estimation is imperfect in practice, we also propose a prior-guided semi-blind deconvolution method. Numerical experiments on simulated images with microtubule-like structures have shown that our proposed algorithms can recover filamentous features with high accuracy and outperforms other state-of-the-art deconvolution methods.