Fluorescence correlation spectroscopy (FCS) is a flexible and widely used tool routinely exploited in vivo. While FCS provides estimates of dynamical quantities, such as diffusion coefficients, it demands high molecular concentrations, high signal to noise ratio and long time traces, typically in the minute range. This is in contrast to data acquisition which, in principle, provides information on microsecond timescales that has, thus far, evaded analysis. To overcome these limitations, we adapt novel tools inspired by Bayesian non-parametrics. Within this principled and generalizable framework, which starts from the direct analysis of photon counts, we achieve the same accuracy as FCS, but with much shorter traces, even at nearly single molecule concentrations. Our new analysis extends the capability of confocal microscopy based approaches, by (i) probing processes several orders of magnitude faster in time; (ii) reducing phototoxic effects on living samples induced by long periods of light exposure; (iii) tracking instantaneous molecules concentration; (iv) estimating the molecular and background photon emission rates; and (v) incorporating demanding illumination profiles, such as those arising in two-photon excitation and TIRF microscopy or even Airy patterns by changing the specified point spread function.
Read full abstract