MusiJ: an ImageJ plugin for video nanoscopy.

Sebastian Acuña,Florian Ströhl,Ida S Opstad,Krishna Agarwal,Balpreet S Ahluwalia

doi:10.1364/boe.382735

Abstract

We present an open-source implementation of the fluctuation-based nanoscopy method MUSICAL for ImageJ. This implementation improves the algorithm's computational efficiency and takes advantage of multi-threading to provide orders of magnitude faster reconstructions than the original MATLAB implementation. In addition, the plugin is capable of generating super-resolution videos from large stacks of time-lapse images via an interleaved reconstruction, thus enabling easy-to-use multi-color super-resolution imaging of dynamic systems.

Highlights

The past two decades have witnessed a huge development in nanoscopy techniques that allow to surpass the resolution limit of optical microscopy and provide super-resolution[1]
Our implementation works with the single-precision floating-point or binary32 format which halves the memory usage and speeds up individual computation steps. In principle, this comes at the cost of numerical precision, we found no noticeable difference in image quality between the outputs generated by the two data types in practice
The MATLAB version was obtained from the official MUSICAL website and executed using MATLAB version R2018b

Summary

Introduction

The past two decades have witnessed a huge development in nanoscopy techniques that allow to surpass the resolution limit of optical microscopy and provide super-resolution[1]. A notable exception are 3D SMLM techniques with purely computational 3D information extraction based on aberrations in the microscope’s point spread function (PSF) - these are available in Fiji [16] Despite their impressive resolution, SMLM reconstruction algorithms require data sets comprising thousands. The window may contain data from the trailing part of the PSFs of fluorophores completely outside the window This non-reliability is suppressed in MUSICAL by using a soft window function on the measurement in the window as well as the PSF of candidate locations of emitters (i.e. the test points). The distance function ds is the projection of the expected output vector for a candidate target point on the eigenvectors with non-zero eigenvalues Introducing this distance is equivalent to stitching the reconstructed images based on the energy contributed by the test point in the numerical space of measurements. For a successful translation into a handy tool, we have developed MusiJ, a plugin for Fiji that improves both on the front end and back end of the original MATLAB implementation in several ways

Back end

Front end

Results and discussion

Conclusion

Disclosures