Super-resolution structured illumination in optically thick specimens without fluorescent tagging

Abstract

This research extends the work of Hoffman et al. to provide both sectioning and super-resolution using random patterns within thick specimens. Two methods of processing structured illumination in reflectance have been developed without the need for a priori knowledge of either the optical system or the modulation patterns. We explore the use of two deconvolution algorithms that assume either Gaussian or sparse priors. This paper will show that while both methods accomplish their intended objective, the sparse priors method provides superior resolution and contrast against all tested targets, providing anywhere from ∼1.6× to ∼2× resolution enhancement. The methods developed here can reasonably be implemented to work without a priori knowledge about the patterns or point spread function. Further, all experiments are run using an incoherent light source, unknown random modulation patterns, and without the use of fluorescent tagging. These additional modifications are challenging, but the generalization of these methods makes them prime candidates for clinical application, providing super-resolved noninvasive sectioning in vivo.

Jointly super-resolved and optically sectioned Bayesian reconstruction method for structured illumination microscopy

“…This enables us to minimize noise amplification using the Bayesian framework [7]. Secondly, unlike the 2D SIM approaches, which achieve both OS and SR by using ad hoc processing such as frequency component weighting [8][9][10][11] or spectral merging [12,13], our proposed method intrinsically enables both OS and SR. This is performed by explicitly considering the 3D nature of the object in the imaging model, even though the data are 2D, as in [14].…”