Assessing oxidative stress in retinal tissue using multispectral autofluorescence imaging.

PURPOSE: Oxidative stress plays a central role in retinal degeneration, yet in-vivo assessment remains challenging due to the reliance on invasive or exogenous techniques. Here, we investigate the use of multispectral autofluorescence imaging as a label-free approach to detect stress-induced metabolic alterations in retinal tissue. METHODS: Rodent retinal explants were exposed to mitochondrial (rotenone), endoplasmic reticulum (thapsigargin), oxidative (hydrogen peroxide), excitotoxic (N-methyl-D-aspartate), and glycolytic (iodoacetic acid) stress. Superoxide levels were independently verified using dihydroethidium staining. Autofluorescence images were acquired across 33 excitation-emission channels. RESULTS: Stressor-specific changes in NAD(P)H- and flavin-associated autofluorescence were observed only for rotenone and IAA treatment, resulting in distinct optical redox ratio (ORR) responses. Despite these limited changes in individual fluorophore-associated signals, classification models based on autofluorescence features achieved strong discrimination for selected stressors (e.g., rotenone: ROC-AUC = 1.00±0.00), with reduced performance when all stressors were combined, reflecting heterogeneous responses. CONCLUSION: These findings demonstrate that multispectral autofluorescence imaging captures stressor-specific signatures rather than a uniform oxidative stress phenotype. Importantly, robust classification performance using UV-free channels highlights the potential for clinical translation. Multispectral autofluorescence imaging therefore represents a promising non-invasive tool for assessing oxidative stress and metabolic dysfunction in retinal tissue.