Removal of subsurface fluorescence in cryo-imaging using deconvolution

Ganapathy Krishnamurthi,Grant Steyer,Charlie Y Wang,David L Wilson

doi:10.1364/oe.18.022324

Abstract

We compared image restoration methods [Richardson-Lucy (RL), Wiener, and Next-image] with measured “scatter” point-spread-functions, for removing subsurface fluorescence from section-and-image cryo-image volumes. All methods removed haze, delineated single cells from clusters, and improved visualization, but RL best represented structures. Contrast-to-noise and contrast-to-background improvement from RL and Wiener were comparable and 35% better than Next-image. Concerning detection of labeled cells, ROC analyses showed RL ≈Wiener > Next-image >> no processing. Next-image was faster than other methods and less prone to image processing artifacts. RL is recommended for the best restoration of the shape and size of fluorescent structures.

Highlights

We are developing cryo-imaging technology and applying it to a very large number of biomedical and biotechnology applications, including stem cell therapy, metastatic cancer, gene expression mapping, mouse phenotyping, imaging agents, etc [1,2,3,4,5]
Similar limitations on field of view are found with other optical sectioning approaches such as computational optical sectioning with a high numerical aperture microscope or structured illumination [9,10], more exotic solutions exist with larger fields of view [11]
We address the problem of subsurface fluorescence using image deconvolution methods which use more than two images, presumably providing improved contrast to noise characteristics as compared to Next-image processing

Summary

Introduction

We are developing cryo-imaging technology and applying it to a very large number of biomedical and biotechnology applications, including stem cell therapy, metastatic cancer, gene expression mapping, mouse phenotyping, imaging agents, etc [1,2,3,4,5]. By alternately sectioning and imaging, a cryo-imaging system will acquire 3D, high-resolution, large field of view, color, and molecular fluorescence image volumes from sequential images of the tissue block face. In the case of bright fluorescent cells and transparent tissues, light scatter and subsurface fluorescence is an issue in block face imaging. Incorporation of confocal or two photon microscopes is a powerful approach, the field of view is necessarily limited by a relatively high numerical aperture. To achieve optical sectioning

Methods

Results

Conclusion