Robin Hood: non-fitting, non-smoothing image detrending for bleaching correction

Abstract

Recent advances in protein labelling--gene tagging with CRIPSR-Cas9--have made it possible to label proteins of interest endogenously. This represents a major breakthrough in the field of quantitative microscopy, especially when quantifying protein-protein interactions. This is because over-expression of labelled proteins may cause a distortion in localization, function and perhaps artificially force protein-protein interactions due to crowding effects. A microscopy technique that is particularly well suited to detect protein interactions with low photon budgets is number and brightness (N&B). Detrending (removal of global trends in data) is a necessary pre-processing step to N&B calculations, but all current detrending methods perform poorly at low intensities. Here, we present the Robin Hood automatic detrending algorithm which performs well at low intensities, evaluating it with simulated and low photon budget live cell images. RH is available as an ImageJ plugin and as an R package.\n\nSTATEMENT OF SIGNIFICANCEFluorescence microscopy in general and Fluorescence fluctuation methods in particular are very much dependent on detrend algorithms and so far, the user needs to decide an arbitrary number to correct for bleaching when using the box plot or the running average approach. Here, we have developed a tool available to everybody as an ImageJ plugin that is automatic, user-free and able to correct bleached images with very low photon counts.