I have been looking into this issue for a couple of years now. I first
would say that 200-300nm is going to be tough with a high n.a. objective.
High NA objectives tend to have working distances of about 170um under
the coverslip. Also, don't cheat and use a #1 coverslip. The spherical
aberations will kill you. But assuming that one could actually see that
far into the tissue with a fairly high NA objective....
 
These findings were presented at Scanning '96 and I am working on paper
on this topic. In general, there are no standard tests that have been
accepted to evaluate this sort of thing. I've created a couple to look at
resolution and linearity (how quantifiable is the data). In our tests
comparing the resolution of the Deltavision deconvolution system (Applied
Precision, Inc.) and a Biorad MRC600 (same optics on both systems) and
the new Leica system show that our MRC does a little worse than the
Rayleigh limit (about 250 nm lateral and 500 nm axial), the Leica (I
can't remember the name of their new system) did the Rayleigh limit (225
nm lateral, 450 axial) and the the undeconvolved also did the Rayleigh
limit. After deconvolution, (measured PSF, 15-iterations of the
reiiterative-constrained method of Sedat and Agard), we get about 90nm
lateral and 400 axial. We suspect that the resolution limit that we are
hitting in deconvolution is related to the voxel size and we will soon be
trying to push it even further.
 
In terms of linearity, we measured a mixed population of varible
intensity beads suspended in a 3-D matrix (a Gelvatol, DABCO mixture).
Again the same optics on both systems but only on the MRC600 and the
Deltavision system. The MRC600 did a poor job of measuring the intensity
of the beads, even over the first log of dilution. The predeconvolved
data from Deltavision did better (about 2 logs of good measurement) and
the deconvolved DV data was the best, it was successful at measuring the
full 2.5 log range of the beads within the mixed population.
 
The sensitivity requires a different kind of target that we are working
on creating. Emperically, it is 50 - 100 times more sensitive for
equivalent exposures.
 
Wavelength seperation is much better because excitation and emmissions
can be optimized for the particular fluorochrome.
 
Photobleaching is noticably less. While hard to measure, Coue,et al. have
shown that photodamage is proportional to flux-density and Amos and White
have shown that the flux density for confocal is as much as 25,000
greater than equivalent power with traditional light as in deconvolution.
 
Unless people really want to get into flame wars, lets try to keep
responses off of the listserv.
 
________________________________________________________________________________
 
 
Paul Goodwin
Image Analysis Lab
FHCRC, Seattle, WA
 
On Wed, 29 May 1996, Dennis McKearin wrote:
 
>         I am not a subscriber to the Discussion group - please direct your
> responses directly to my e-mail address.
>         I am interested in learning the pro & cons of confocal vs
> deconvolution methods of microscopy. Specifically, I wish to know how the
> two methods compare for double and triple labeling when stained samples are
> 200-300 microns thick. What limitations exist between the two techniques -
> do both provide images of equal resolution and sensitivity or is one
> superior, especially as applied to multiple labeling experiments? Are
> sample bleaching rates equal?
>         Thank you for your time and help.
>
> Sincerely,   Dennis McKearin
>