Re: colocalization between confocal and two-photon: voxel matching

Posted by Andreas Bruckbauer on
URL: http://confocal-microscopy-list.275.s1.nabble.com/xyzt-image-tp7580102p7580116.html

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
*****


 Hi Arvydas,

make sure that the two-photon laser is aligned well with the visible laser, otherwise there will be a shift between the two channels, this is usually done during service of the microscope but quite often not with the best accuracy. For the optical slice of the two-photon signal it is important how large the laser spot at the  back-aperture of the objective is, this can vary between setups and objectives (overfilling or underfilling) Best take z-stacks of some small beads and check how well the channels are aligned in x,y and z and the real size of the sections. You can then discuss the results with the microscope vendor or someone who knows how to set up such a system.

best wishes

Andreas


 

 

-----Original Message-----
From: Julio Vazquez <[hidden email]>
To: CONFOCALMICROSCOPY <[hidden email]>
Sent: Wed, 3 Apr 2013 19:14
Subject: Re: colocalization between confocal and two-photon: voxel matching


*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
*****

Hi Arvydas,

You want to keep your pinhole fully open for two-photon imaging..... in two
photon mode, you are exciting one single spot in the specimen, so there is no
need to have a pinhole to exclude out of focus light, since there isn't any. You
want the pinhole open to maximize signal collection. Do not trust the estimate
of the optical slice thickness in two photon mode that the LSM software gives
you. The voxel size for two photon excitation is defined by the wavelength you
use for excitation, I'm guessing around 800 nm, so the radius of your laser spot
is only 1.6 times greater for DAPI compared to excitation with 488 nm. The
thickness of the optical slice is fixed for two photon (size of exciting laser
spot). You can open the pinhole a little bit for the green chanel to try to
match (I'm guessing to maybe 1.5 Airy units), so this would give you voxels that
are not so different.

Keep in mind that the accuracy of your colocalization will depend on other
factors: how you threshold, how you subtract background, amount of noise in your
images, etc.... If you are trying to "colocalize" structures that are
significantly larger than the resolution of you microscope, the error
contributed by the differences in voxel size will be comparatively small. If the
structures you are comparing are near or below the resolution limit, then the
error will get comparatively higher. As an extreme example, if you have two
single molecules 100 nm apart, their images will mostly overlap and give a high
degree of colocalization using conventional colocalization methods, while in
reality the two molecules do not overlap at all. So my advice is to adjust teh
pinhole for 488 to try to match the voxel size for DAPI as best as you can, and
then use a sound colocalization method (and good controls) for your
colocalization.

Deconvolving the data might be a good idea to reduce noise and tighten your PSFs
before running your analysis.

I found the discussions on colocalization provided by SVI (scientific volume
imaging) to be very useful. You can Google colocalization + SVI. to access their
documentation.

Just saw Brian's response, so some of the above is redundant.

--
Julio Vazquez
Fred Hutchinson Cancer Research Center
Seattle, WA

http://www.fhcrc.org



On Apr 3, 2013, at 10:28 AM, Arvydas Matiukas wrote:

> *****
> To join, leave or search the confocal microscopy listserv, go to:
> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
> *****
>
> Hello list,
>
>
> One of my Core users wants to measure colocalization between
> DAPI  labeled and  FITC-immunolabeled structures in nuclei of
> her fixed cells. Visual inspection shows good level of colocalization.
>
> However, what would be the correct way to quantify it.
> Our confocal is LSM510 META NLO. No UV or 405nm laser
> so two-photon mode at 780nm  is used to excite blue dye
> while FITC is excited at 488nm in confocal mode . At default
> pinhole settings  (1 airy unit for confocal and max=~10 A.u.
> for two-photon)  both blue and green signals are good
> but the optical slices (and voxel sizes)  differ by the factor of 10.
> Letting the software (LSM)  to equalize the slices at 0.6um reduces the
> pinhole to 0.76 for confocal and 1.0 for two-photon.
> Unfortunately at such small pinhole for two-photon the
> blue signal is lost in noise (even at 5% laser power and 16 times
> averaging, further increase may bleach the dye during 15 min
> long scan).  I  can get the blue signal back to a measurable level
> by increasing  the two-photon pinhole to 4 A.u. but this creates
> a four-fold mismatch in blue/green voxel size. I could boost the
> blue signal by increasing pixel size but the loss in resolution
> is not desirable.
>
> My first thought was to switch to  red DNA stain and then
> do colocalization between two confocal signals with identical
> voxels. However, the user already completed experiment
> using  DAPI staining, and it worked well. My question is
> what would be the correct way to acquire the confocal/two-photon
> z-stacks and subsequently quantify the colocalization.
> And related question if reducing pinhole works the same
> way for two-photon as for a regular confocal.
>
> Any feedback/advice/idea are very welcome,
> Arvydas
> ----------------------------
>
>
>
> Arvydas Matiukas, Ph.D.
> Director of Confocal&Two-Photon Core
> Department of Neurosci& Physiology
> SUNY Upstate Medical University
> 766 Irving Ave., WH 3167
> Syracuse, NY 13210
> tel.: [phone]
> fax: [phone]
> email: [hidden email]