From FRET Imaging to Practical Methodology for Kinase Activity Sensing in Living Cells - Progress in Molecular Biology and Translational Science - page 181

Regardless of the fitting method employed (and as previously described),
the final purpose of the FLIM data analysis is to correctly extract both the
proportions and the lifetime components for each pixel of the image. This
can be achieved preferentially without constraining any fitting parameters
(cf. Fig. 5.18 ). However, for sample-emitted fluorescence whose intensity
decay is multiexponential, the correct estimation of all parameters with a
standard fitting method requires a large number of photons

100,000 85 ). For decreasing this number and consequently the acquisi-
tion time and phototoxicity, we have already detailed in previous paragraphs
two solutions based on algorithm implementation (MLE and AMDI). An
alternative solution consists in reducing the number of fitting parameters
by constraining one of them, for example, the donor lifetime, but it makes
it necessary that the donor lifetime is monoexponential and that it has been
precisely determined in a previous experiment (which requires a modified
biosensor without acceptor). Another possibility is to analyze the data glob-
ally, but it is valid only if one can assume the lifetime information to be iden-
tical for all pixels of the FLIM image (or a selected area). 86,87 In all cases, it
requires expertise and computation time for obtaining reliable lifetime
values with fitting methods.
Once the lifetime components have been correctly estimated, the
resulting lifetime images are displayed using a color scale and are usually
overlaid on the intensity image for weighting the lifetime value with the
fluorescence intensity (cf. Fig. 5.18 ). However, due to the large number
of parameters (proportion and lifetime components), it is not possible to rep-
resent the evolution of all of them. In order to simplify data representation,
the mean lifetime t mean defined by Eq. (5.13) has been largely used in FRET
experiments. 8-11 However, the mean lifetime does not correspond to the
correct average lifetime, which is defined by Eq. (5.12) . This average
lifetime has been recently used for biosensor activity recording. 88
However, the reader should be informed that this quantity is not
monotonous as a function of the donor lifetime in the presence of the
acceptor. In other words, one average lifetime value can be found with
two distinct donor lifetimes in the presence of the acceptor for a fixed
proportion of the interacting donor. Consequently, the knowledge of
both the average lifetime and the fraction of the interacting donor is not
enough to entirely characterize the fluorescent sample. Due to the
nonuniqueness of the average lifetime, it is also necessary to combine it
with another parameter (e.g., FRET efficiency).

( N >