Direct Photon-by-photon Analysis of Time-Resolved Pulsed Excitation Data using Bayesian Nonparametrics

Abstract

AbstractLifetimes of chemical species are typically estimated, across each illuminated spot of a sample, by either fitting time correlated single photon counting (TCSPC) decay histograms or, more recently, through phasor analysis from time-resolved photon arrivals. While both methods yield lifetimes in a computationally efficient manner, the performance of both methods is limited by the choices made when fitting a TCSPC histogram. In addition, phasor analysis also requires setting the number of chemical species by hand before lifetimes can be determined. Yet the number of species itself is encoded in the photon arrival times collected for each illuminated spot and need not be set by hand a priori. Here we propose a direct photo-by-photon analysis of data drawn from pulsed excitation experiments to infer, simultaneously and self-consistently, the number of species and their associated lifetimes from as little as a few thousand photons for two species. We do so by leveraging new mathematical tools within the Bayesian nonparametric (BNP) paradigm that we have previously exploited in the analysis of single photon arrivals from single spot confocal microscopy. We benchmark our method on simulated as well as experimental data for one, two, three, and four species with data sets from both immobilized and freely diffusing molecules at the level of one illuminated spot.SUMMARYPhoton arrivals obtained from fluorescence experiments encode not only the lifetimes of chemical species but also the number of chemical species involved in the experiment. Traditional methods of analysis, such as phasor methods and methods relying on maximum likelihood or (parametric) Bayesian analysis of photon arrivals or photon arrival histograms of TCSPC data, must first ascertain the number of chemical species separately and, once specified, determine their associated lifetimes. Here we develop a method to learn the number of fluorescence species and their associated lifetimes simultaneously. We achieve this by exploiting Bayesian nonparametrics. We benchmark our approach on both simulated and experimental data for one species and mixtures of two to four species.