Super Resolution Image Reconstruction for Structured Illumination Microscopy (SIM) as ImageJ Plugin

Roland Nussbaumer

Semester Project

Section of Microengineering, EPFL

January 2015

Abstract

All optical microscopy systems are physically limited by diraction. Since this limit (200nm in optimal conditions) is often too constraining for biomedical imaging, researchers began to look for super resolution imaging techniques. Structured Illumination Microscopy (SIM) is one widely used method to exceed that resolution limit. The idea is to use spatially struc- tured light to illuminate the object structure, which results in Moire-fringes of lower frequency, now detectable through the optical system. Using dig- ital reconstruction algorithms, the additional information hidden in those fringes can then be extracted.

The method was was proposed in 2000 and then several manufacturers like Nikon and Zeiss produced microscopy systems which are based on SIM. But they provide their software only to licensed users. The aim of this project was, for the rst time to our best knowledge, the creation of an open source ImageJ package which we named SIMage. By illuminating the object with a sinusoidally structured pattern, one can reach twice the resolution as in conventional wide-eld microscopy. We have reached that limit with simulated data. However, by applying the algorithm on real data, it has become clear that a precise estimation of the pattern parameters and the characteristics of the optical system is crucial in order to obtain the maximum resolution improvement.