[FLIMfit-users] SymphoTime vs. FLIMfit

Munro, Ian i.munro at imperial.ac.uk
Fri Mar 10 14:52:52 GMT 2017


Hi Marko

There is a general problem, in minimisation, of false minima or multiple minima.
Minimisation software simply searches a ( in your case 4D) surface looking for the lowest value of some figure-of-merit (e.g. chi-squared).
A 2D analogy would be trying to find the lowest point in a landscape.
If there are 2 equally deep valleys  (2 minima) either side of a ridge then your result can depend on which side of the ridge you start your search.
From what you say, you are taking an approach, referred to in FLIMfit as ‘Global binning’.
The FLIMfit software offers you the alternative of ‘Global fitting’ where, rather than fixing one lifetime in advance, you fit both but tell the software to assume that the 2 lifetimes are constant across the image.
Maybe this approach would shed some light on your problem.

For IRF’s you need something with a known response & ideally emitting at the same wavelength as your sample.
The gold-nanorods are great but IIRC only appropriate for two-photon excitation.
Alternatively you can use a dye with either a  very short lifetime or a mono-exponential decay and a known lifetime (the 'Reference deconvolution’ option in ‘IRF type’. You can also use a scattering sample and the scattered excitation light although you need to be careful as  the different wavelength may introduce a time-shift and this will then require you to shift the IRF data to compensate.

Best Wishes

Ian


> On 8 Mar 2017, at 13:16, Marko Šoštar <markosostar at gmail.com> wrote:
> 
> Hi,
> my name is Marko and I am a phd student on the Institute Ruđer Bošković in Zagreb, Croatia. 
> I recently did some FLIM-FRET measurements where I tried to find out FRET-ing population in a sample by fitting on the double-exponential model N(t) = A1*exp(−t/t FRET) + A2*exp(−t/t 0). Ideally, I would measure donor lifetime (t0) from donor only sample (mono-exponential decay), and then use this as a fixed parameter in subsequent fitting procedures (with acceptor present). And, that worked fine until recently, when I came across a sample where very small change in donor lifetime (fixed parameter-I was playing with manually changing it) would yield substantial difference in fitted A1 and A2, with equally good fit quality (chi-squared test). So, now I can't decide which amplitudes to use. Does anybody have some advice how to resolve this issue?
> We use Leica confocal microscope paired with PicoQuant TCSPC system and SyphoTime software for analysis. I should mention that I never measured IRF (software offers to approximate it), so I'm not sure how this affects fitting results. Also, I recently became aware of the FLIM fit software, and I was wondering is there anyone experienced with both SymphoTime and FLIM fit. What would be the better choice? Could this fitting issue be resolved by using FLIM Fit (maybe it has some extra features for assessing goodness of fit?)? Also, could you tell me what would be the best sample (way) for measuring the IRF function (gold nanorods?- where to get them?).
> I would greatly appreciate any advice.
> 
> Thank you and kind regards,
> Marko Šoštar
> _______________________________________________
> FLIMfit-users mailing list
> FLIMfit-users at lists.openmicroscopy.org.uk
> http://lists.openmicroscopy.org.uk/mailman/listinfo/flimfit-users



More information about the FLIMfit-users mailing list