AFid: A tool for automated identification and exclusion of autofluorescent objects from microscopy images

This upload contains all of the raw data and scripts needed to reproduce all the Figures in Baharlou,Canete et al., 2020. The manuscript is currently in review with a preliminary version available on bioRxiv (https://doi.org/10.1101/566315). 

Abstract:

Autofluorescence is a long-standing problem that has hindered the analysis of images of tissues acquired
by fluorescence microscopy. Current approaches to mitigate autofluorescence in tissue are lab-based and involve either
chemical treatment of sections or specialized instrumentation and software to ‘unmix’ autofluorescent signals. Importantly,
these approaches are pre-emptive and there are currently no methods to deal with autofluorescence in acquired
fluorescence microscopy images.To address this, we developed Autofluorescence Identifier (AFid). AFid identifies autofluorescent pixels as discrete objects in multi-channel images post acquisition. These objects can then be tagged for exclusion from downstream
analysis. We validated AFid using images of FFPE human colorectal tissue stained for common immune markers. Further, we demonstrate its utility for image analysis where its implementation allows the accurate measurement of HIV-Dendritic Cell interactions in a colorectal explant model of HIV transmission Therefore, AFid represents a major leap forward in the extraction of useful data from images plagued by autofluorescence by offering an approach that is easily incorporated into existing workflows and that can be used with various samples, staining panels and image acquisition methods. We have implemented AFid in ImageJ, Matlab and R to accommodate the diverse image analysis community.