DeepMetaboFLIM – Deep learning methods for the Metabolic characterization of brain function and aging by label-free Fluorescence Lifetime IMaging

Abnormal energy metabolism is central to many pathologies, including cancer or neurodegenerative diseases. Early diagnosis calls for the development of minimally invasive and sensitive techniques to monitor the metabolic state of tissues. The label-free fluorescence lifetime imaging (FLIM) of the cellular metabolic cofactors NADH and FAD holds great potential in this aim. Yet, the limited sensitivity and discrimination ability of the current analysis methods of these complex data have prevented the dissemination of this technique.


This proof-of-concept project will implement dedicated deep learning analysis procedures to 3D FLIM metabolic imaging of the Drosophila brain, to establish that this new methodology allows discriminating subtle contribution of specific metabolic pathways. The sensitivity of this method to physiologically-relevant conditions will be demonstrated through (i) the elucidation of metabolic regulations underlying memory formation, and (ii) the construction a 4D (3D+time) metabolic map of brain aging.