To: Multiple recipients of list CONFOCAL <[log in to unmask]>
In-Reply-To:  -Michael  [log in to unmask]
 
 >A specific question we have is which confocal has the best Z axis
 >resolution.  Using VoxelView we have been reconstructing volumes from
 >serial sections collected with a 60X objective at approximately a 2X
 >zoom with a BioRad MRC-600.  So, for instance, would the new Nikon or Noran
 >microscope provide images with high enough resolution to do this
 >reconstruction?
 
Our BioRad MRC 600 gives 0.5um Z resolution in reflection mode, tested with
the mirror slide test, using a x63 NA 1.4 Zeiss Planapochromat.  I can see
no reason why any other fully confocal system with an adjustable pinhole and
adequate provision for alignment shouldn't give just the same Z resolution
since it is a function of NA (squared) and (to a certain extent) of pinhole
size.  However systems which use slit-scanning (Meridian, Viewscan) or are
optically equivalent to slit scanning (Noran, Lasertec) would be worse, and
so would spinning-disk systems since they have larger than optimal pinholes.
 
I strongly recommend the mirror slide as the only Z resolution test that is
a real absolute standard.  (There is no such thing as an infinitely small
fluorescent bead).  It is also quick and simple to carry out if one is
evaluating a microscope.  (Provided reflection facilities are available -
the Olympus often seems to ship eqipped only for fluorescence).
 
The other important point is that the Z-drive itself should be capable
of the required resolution (minimum step no larger than 0.2um) - on a
Zeiss Axiophot the Biorad motor ony just makes it (min. step 0.18um).
 
Those who seek to improve the appearance of VoxelView reconstructions
might be interested in an extremely simple image-processing technique
described in a paper by yours truly & Colin Sheppard in BioImaging
(scheduled to come out in July but I haven't seen it yet).  Basic (very
basic) source code is included.  We also presented it at the February
confocal conference.  Anyone who wants to play with it (at their own
risk!) and would like source code or a very experimental executable
is welcome to contact me.
 
Finally I would like to echo Alan Hibbs' comments, that resolution (in
any direction) is only one factor, and sensitivity is far more likely to
vary significantly between systems than resolution.
 
                Guy Cox
                [log in to unmask]