ABSTRACT
We demonstrate a proof of concept for detecting heterogeneities and estimating lifetimes in time-correlated single-photon-counting (TCSPC) data when photon counts per molecule are low. In this approach photons are classified as either prompt or delayed according to their arrival times relative to an arbitrarily chosen time gate. Under conditions in which the maximum likelihood (ML) methods fail to distinguish between heterogeneous and homogeneous data sets, histograms of the number of prompt photons from many molecules are analyzed to identify heterogeneities, estimate the contributing fluorescence lifetimes, and determine the relative amplitudes of the fluorescence, scatter, and background components of the signal. The uncertainty of the lifetime estimate is calculated to be larger than but comparable to the uncertainty in ML estimates of single lifetime data made with similar total photon counts. Increased uncertainty and systematic errors in lifetime estimates are observed when the temporal profile of the lifetime decay is similar to either the background or scatter signals, primarily due to error in estimating the amplitudes of the various signal components. Unlike ML methods, which can fail to converge on a solution for a given molecule, this approach does not discard any data, thus reducing the potential for introducing a bias into the results.
