Hidden in visible light: spectral-temporal unmixing of lung tissue autofluorescence in a fibre-based system

Autofluorescence molecules, including metabolites and structural tissue components that play a crucial role in cellular processes, are often disrupted during lung cancer oncogenesis. These fluorophores emit photons when they return to their ground state after excitation, where the average time spent in the excited state, i.e., fluorescence lifetime, is influenced by their environment. The fluorescence process is highly sensitive to nuanced environmental changes, making it an ideal method for personalized lung cancer diagnosis. However, current fibre-based fluorescence devices typically use broad wavelength channels, and computational methods often compute average fluorescence lifetimes, reducing the sensitivity and specificity for detecting subtle disease changes. We provide evidence that a high-resolution spectral-temporal time-resolved fluorescence spectroscopy (TRFS) device (0.5 nm, 50 ps) coupled with the multichannel fluorescence lifetime estimation (MuFLE) model can unmix underlying individual components label-free using their spectral and temporal characteristics simultaneously. In lung tissue ex vivo, we extract paired fluorescence lifetimes and emission profiles, and spatially correlate these measurements with endogenous fluorophores using confocal FLIM and complementary antibody staining. This technique promises label-free real-time tracking of endogenous fluorophores during lung cancer diagnosis, offering a more personalized, precise metabolic and structural assessment during bronchoscopy in real time.