Building Fluorescence Lifetime Maps Photon-by-Photon by Leveraging Spatial Correlations.

Area of Science:

• Biophysics
• Microscopy
• Computational Biology

Background:

• Fluorescence lifetime imaging microscopy (FLIM) is crucial for quantitative subcellular analysis.
• Current FLIM methods face challenges in spatial resolution, absolute lifetime mapping, species identification, and uncertainty propagation.
• Existing techniques often analyze pixels independently, limiting spatial resolution and failing to provide absolute lifetime maps.

Purpose of the Study:

• To develop a novel, self-consistent framework to address the limitations of current quantitative FLIM analyses.
• To simultaneously learn the number of fluorophore species and their absolute lifetime maps.
• To improve spatial resolution and handle low photon budgets in FLIM data.

Main Methods:

• A doubly nonparametric Bayesian framework is proposed.
• Beta-Bernoulli process priors are used to learn the number of contributing fluorophore species.
• Gaussian process priors are employed to generate absolute lifetime maps, utilizing information from all detected photons.

Main Results:

• The framework successfully overcomes key challenges in quantitative FLIM analysis.
• It enables the discrimination of fluorophore species with lifetime differences as small as 0.3 ns.
• Robust performance was demonstrated across diverse synthetic and experimental datasets, even with as few as 1000 photons.

Conclusions:

• The proposed doubly nonparametric framework offers a significant advancement for quantitative FLIM.
• This method provides accurate absolute lifetime maps and species identification, enhancing subcellular environment characterization.
• The framework's ability to perform robustly with low photon counts opens new possibilities for FLIM applications.