Posted by
Jules Girard on
URL: http://confocal-microscopy-list.275.s1.nabble.com/Re-Using-a-mirror-for-axial-resolution-testing-tp6619710p7581106.html
*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy*****
Dear all,
I've been following that interesting discussion with a bit of delay. Although
most has been already said, I feel that a clearer answer can be given to the
initial question (I was about to say more concise, but I now figure out that my
contribution is not so short...). Let me try to do so.
Definitions are important here :
-Resolution (axial and lateral), is the capability of an imaging system to resolve small details of an object, which can be quantified in different ways. In Fourier
space, it is related to both the extent (support) and the magnitude of the OTF.
-(Axial) Optical Sectioning is the capability of an imaging system to get rid of
out of focus signal.
In my opinion, the most complete way to quantify resolution is to think in the
frequency space and use OTF, especially when deconvolution is considered
and/or with samples that are not pointillistic (i.e. that have an inhomogeneous
spatial spectrum). But a more common and often convenient way to quantify
resolution is to stay in direct space and use PSF + Rayleigh criterion/FWHM.
This is however an arbitrary and restrictive way of defining/measuring
resolution.
Brad Amos previous answer seems to be confusing axial and lateral resolution.
Unlike what is often thought, lateral resolution is scarcely improved by confocal
microscopy compared to WF (I generally consider that improvement to be
inexistent), as it has been shown by Guy Cox and Colin Sheppard [1]. The
sqrt(2) factor is valid only for very small pinhole and infinite signal : in practice
that improvement is exactly compensated by the signal (over noise ratio)
decrease. The improvement is however clear for axial resolution, as it can be
seen with comparison of 3D OTF in both WF and confocal case (cf. that great
99 paper by Gustafsson [2]). The improvement is much better than a sqrt(2)
factor here, even with a reasonable pinhole size.
You need Optical Sectioning when you have a 'thick' or autofluorescing sample,
to avoid getting your in focus signal drown in blurry background signal. It has
been pointed out already, but let me remind here that it is sometimes possible
to axially resolve two objects even without any sectioning power: in the WF
case, the out of focus signal is quickly spread laterally, and that one can
generally distinguish two beads axially aligned in a z-stack. In Fourier space,
this can be seen as a contribution of "diagonal" frequencies (i.e. with both
lateral and axial direction components) that are out of the dark cone (or
"singularity" as Lutz was calling it) of the WF OTF. Ideally, with a well known
PSF, a confined object and no shot noise increase, one could deconvolve such
blurry 3D image and not care about optical sectioning anymore (except for the
object features that are mainly described with spatial frequencies situated in
that dark cone, i.e. flat layers of fluorophores more or less orthogonal to the z
axis).
To sum up :
-If you care more about resolution, I would say that using small beads is more
appropriate. If your bead is 100nm or smaller, you get a good approximation of
your PSF from your direct measurement and you can use it for whatever
arbitrary criterion you want to use (Rayleigh, FWHM, Sparrow or OTF...).
-If you care more about optical sectioning, then you should use a thin
fluorescent layer : the spreading of the out of focus light will be compensated
by the quasi infinite lateral extent of your sample (within the limit pointed out
by James Pawley) and you should get a I(z) function that is suitable to define
optical sectioning. Alternatively, you could also use a PSF measurement from a
small bead and integrate the signal you get at each z position (i.e. project your
I(x,y,z) measured function onto a I(z) one). The result should theoretically be
the same than with the thin layer, but you will have a much lower signal and
and a lot of pixels to sum when going out of focus, leading to some poor SNR.
Best,
Jules
[1] G. Cox, C.J.R. Sheppard, Practical limits of resolution in confocal and non-
linear microscopy., Microsc. Res. Tech. 63 (2004) 18-22.
[2] M.G.L.
Gustafsson, Extended resolution fluorescence microscopy, Curr. Opin. Struct.
Biol. 9 (1999) 627-628.
--
Jules Girard, PhD
Postdoctoral Researcher
in
Physics of Living Systems group
Department of Physics and Astronomy
VU University Amsterdam
Room U-059
De Boelelaan 1081
1081 HV Amsterdam, The Netherlands
Tel : +31 (0)20 59 879 15
Email:
[hidden email]