Rapid label-free imaging of autofluorescence lifetime of biological samples with low photon counts

Rapid and point-of-care bacterial imaging, detection, and classification in healthcare settings are currently constrained by the prolonged processing times of culture-based methods, which typically require more than three days. Autofluorescence lifetime imaging presents a promising alternative due to its high sensitivity and specificity. However, traditional implementations are often slow and difficult to scale to larger areas, making them costly and impractical for routine use. To overcome these limitations, we adopt a method inspired by recent work [Xiao, et al. Optica 10, 944-951 (2023)], which enables rapid autofluorescence lifetime imaging over large areas (e.g., 5 cm × 5 cm) using low-cost components. This approach combines low-photon count measurements acquired via time-correlated single photon counting with autofluorescence intensity images, processed through deep neural networks to reconstruct high-resolution lifetime maps. Using this strategy, we successfully demonstrate spatial demarcation between regions containing bacteria and the background regions, on biological targets such as wounds and bacteria immobilized on filter membranes.