Can we use rapid lifetime determination for fast, fluorescence lifetime based, metabolic imaging? Precision and accuracy of double-exponential decay measurements with low total counts

Abstract

Fluorescence lifetime imaging microscopy (FLIM) can assess cell’s metabolism through the fluorescence of the co-enzymes NADH and FAD, which exhibit a double-exponential decay, with components related to free and protein-bound conditions. In vivo real time clinical imaging applications demand fast acquisition. As photodamage limits excitation power, this is best achieved using wide-field techniques, like time-gated FLIM, and algorithms that require few images to calculate the decay parameters. The rapid lifetime determination (RLD) algorithm requires only four images to analyze a double-exponential decay. Using computational simulations, we evaluated the accuracy and precision of RLD when measuring endogenous fluorescence lifetimes and metabolic free to protein-bound ratios, for total counts per pixel (TC) lower than 10 4 . The simulations were based on a time-gated FLIM instrument, accounting for its instrument response function, gain and noise. While the optimal acquisition setting depends on the values being measured, the accuracy of the free to protein-bound ratio α 2 /α 1 is stable for low gains and gate separations larger than 1000 ps, while its precision is almost constant for gate separations between 1500 and 2500 ps. For the gate separations and free to protein-bound ratios considered, the accuracy error can be as high as 30% and the precision error can reach 60%. Precision errors lower than 10% cannot be obtained. The best performance occurs for low camera gains and gate separations near 1800 ps. When considering the narrow physiological ranges for the free to protein-bound ratio, the precision errors can be confined to an interval between 10% and 20%. RLD is a valid option when for real time FLIM. The simulations and methodology presented here can be applied to any time-gated FLIM instrument and are useful to obtain the accuracy and precision limits for RLD in the demanding conditions of TC lower than 10 4 .

A Review of New High-Throughput Methods Designed for Fluorescence Lifetime Sensing From Cells and Tissues

“…Analogous to integral approximation methods in calculus, having more time gates with smaller widths results in higher lifetime accuracy, particularly for fluorophores with short lifetimes or multiexponential decay, while having a longer overall detection time improves accuracy for longer fluorescence lifetimes [ 46 ]. Adding more gates or increasing the detection window size leads to longer acquisition times and lower throughput; however, analysis algorithms [ 21 , 47 ] and even deep learning [ 22 ] have been implemented to maintain accuracy while collecting less images or gates. Similar detectors used for TCSPC can be used in time-gated FLIM, including modified SPAD arrays with gating electronics.…”

“…The applications of fluorescence lifetime measurements range significantly. Some examples include separating fluorophores that have similar spectra but different lifetimes [4][5][6][7][8][9][10][11][12][13], exploiting the difference in lifetime of free and bound metabolites for metabolic study and diagnosis [14][15][16][17][18][19][20][21][22][23][24], using FRET to screen for binding or inhibition of exogenous and endogenous molecules of interest [2,5,[25][26][27][28][29][30][31][32][33][34][35], evaluating the conformation and stability of proteins or other molecules in varying environments [31,36], or even using lifetime as an additional parameter in fluorescent inks for anti-counterfeiting [37], though this list is far from exhaustive. New methods and technologies continue to be discovered, further establishing the wide-reaching significance and potential of fluorescence lifetime.…”

Using in vivo multiphoton fluorescence lifetime imaging to unravel disease-specific changes in the liver redox state

“…This work has also used Becker & Hickl TCSPC data recording on a Dermainspect MPM in our in vivo real time clinical imaging approach. Others have proposed a wide-field, timegated FLIM approach based on only four images, with high precision being achieved by either a summing of frames or by use of pixel-binning [77]. However, our work here has shown that spatial resolution may be compromised by pixel binning, leading to erroneous results in heterogeneous samples and the summing of frames may also need to be corrected for intravital motion [10].…”

Fast algorithm for time decay estimation with applications to electrostatic ion beam traps

“…In such experiments, fast estimation of the lifetime is an important feature. The problem of rapid evaluation of lifetime is important not only for ion traps, but also for luminescence and fluorescence applications [22][23][24], and is also related to the estimation of damping sinusoids and exponential parameters, which has received considerable interest in the signal processing community (see for example [25][26][27] and references therein).…”