High-speed Super-resolution Imaging through Interpolated Deconvolution of Live-cell TIRF Images

Abstract

Listen

Translate article

Recent progresses in overcoming the diffraction-limited optical resolution have mostly relied on spatial modulation of fluorophore's distribution between its bright and dark photophysical states. Such feat is often accomplished through complex and expensive experimental setups, and almost all the related super-resolution imaging techniques are hostile towards live-cell studies. In contrast, super-resolution imaging through interpolated deconvolution (Carrington et al., 1995, Science, 268:1483) is a post-acquisition image-processing technique that is independent on microscope platform. Its efficient collection and utilization of fluorescence photons also make the technology potentially a preferred method for live-cell imaging. However, performance of the technique is context-dependent: e.g., weak fluorescence signals and clustered sub-resolution structures typically yield poor deconvolution results. We have evaluated such difficulty using realistically simulated TIRF images of GLUT4 glucose transporters in cultured adipocytes, whose average diameter of 75nm is far below the optical resolution. An essential image-processing step isolating regions of high information content from a TIRF image was discovered, which enables subsequent deconvolution resolution approaching 100nm. Detailed analysis of deconvolution results as a function of signal-to-noise qualities of the original images suggests that super-resolution details can be resolved with TIRF images of live cells acquired at speed up to 10 fps.