Single Photon, Time-Gated, Phasor-based Fluorescence Lifetime Imaging Through Highly Scattering Medium

Abstract

Fluorescence lifetime imaging (FLI) is a powerful tool for in vitro and non-invasive in vivo biomolecular and cellular investigations. Fluorescence lifetime is an intrinsic characteristic of any fluorescent dye which, to some extent, does not depend on excitation intensity and signal level. However, when used in vivo with visible wavelength emitting fluorophores, FLI is complicated by (i) light scattering as well as absorption by tissues, which significantly reduces fluorescence intensity, (ii) tissue autofluorescence (AF), which decreases the signal to noise ratio and (iii) broadening of the decay signal, which can result in incorrect lifetime estimation. Here, we report the use of a large-frame time-gated single-photon avalanche diode (SPAD) imager, SwissSPAD2, with a very short acquisition time (in the milliseconds range) and a wide-field microscopy format. We use the phasor approach to convert each pixel's data into its local lifetime. The phasor transformation provides a simple and fast visual method for lifetime imaging and is particularly suitable for in vivo FLI which suffers from deformation of the fluorescence decay, and makes lifetime extraction by standard fitting challenging. We show, for single dyes, that the phasor cloud distribution (of pixels) increases with decay broadening due to scattering and decreasing fluorescence intensity. Yet, as long as the fluorescence signal is higher than the tissue-like phantom AF, a distinct lifetime can still be clearly identified with an appropriate background correction. Lastly, we demonstrate the detection of few hundred thousand A459 cells expressing the fluorescent protein mCyRFP1 through highly scattering phantom layers, despite significant scattering and the presence of the phantom AF.

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

“…All of the parameters not explicitly varied are held constant across all of the figures. The parameters not varied are held fixed at the following baseline values: lifetime between 1 and 10 ns, which is the typical lifetime range of a fluorophore 18 , 85 ; 2 species, which is most frequent in related studies 18 , 19 , 23 ; and fraction of molecules contributing photons from different species set at 50%:50%.…”

In vitro and in vivo NIR Fluorescence Lifetime Imaging with a time-gated SPAD camera

“…For use with SS2, which requires synchronization with a source signal at around 20 MHz, a frequency divider (TOMBAK, Laser Lab Source Corporation, MT, USA) was used to divide the Mai Tai’s laser repetition rate (∼80 MHz) by a factor of four ( f SYNC = 19.77 MHz reported). SS2 was set to acquire 10-bit gate images consisting of 1,020 accumulated 1-bit gate images, each 1-bit image resulting from exposure of each SPAD pixel to the incoming photon flux for a user-specified duration of n E × 400 ns, where n E is some number typically between 1 and 100 23 . During that 1-bit accumulation period, each SPAD is “on” ( i .…”

“…Integration time for each gate image is calculated following ref. 23 as: T int = (8 n G −1) bNT Sync , where n G is a gate sequence parameter (typically n G = 100), b is the number of 1-bit gate images per final image ( b = 1,020), N is the number of accumulations (e. g. N = 70) and T sync = 1/ f SYNC is the period of the synchronization signal. T int was set between 1.02-4.08 s depending on the sample brightness.…”

“…All phasor analysis was performed using Alligator 18 as detailed elsewhere 23,34 . In brief, for every pixel with coordinate ( x, y ) within the MFLI region of interest, the uncalibrated discrete phasor z ( x, y ) = g ( x, y ) + i s ( x, y ), where i the complex root of −1, was retrieved from the time-gated decay { G p ( x, y )} p = 1,.., G according to Eq.…”

Deep Learning in Biomedical Optics

“…The phasor method is the most widely‐accepted alternative for lifetime representation [87]. However, accurate quantification using the phasor method requires careful calibration, and when considering tissues/turbid‐media in FLI microscopy (FLIM) applications, additional corrections are needed [87,88]. Therefore, it has largely remained qualitative in use.…”