Deep learning based quantification of cutaneous drug uptake from standard epi-fluorescence microscopy

Many topically applied drugs are naturally fluorescent, but quantifying their uptake into skin via fluorescence emission is complicated by both weak fluorescence and often-overwhelming skin autofluorescence. Fluorescence lifetime imaging microscopy (FLIM) has been found capable of identifying and separating the fluorescence of multiple drugs from skin autofluorescence via their different spectral and lifetime properties. This investigation was focused on combining epi-fluorescence microscopy with deep learning for the quantification of naturally fluorescent topically applied drugs. This combination of deep learning and fluorescence imaging may allow for straightforward relative quantification of fluorescent drugs in tissue using only simple, readily available imaging tools.