> Paul,
>
> How does deconvolution solve the lensing problem?  Surely by assuming a
> fixed PSF, they would also produce incorrect results as well.
 
I never claimed that it does, but rather in deconvolution it seems less
severe for the case of multiple beads on top of each other and I mention
it because people get the idea that since it is a confocal, it must be
revealing truth (a confocal wouldn't lie, would it?) The truth of the
matter is that bad optics and non-uniform optical paths through the
specimen effect all imaging. If one thinks that confocal is immune, one is
uninformed. I thought that that was one of Wilson's messages at Scanning.
 
 
> And I thought that you and I had agreed at the Scanning meeting last month
> that it was unfair to use the term "resolution" as a measure ot the spatial
> accuracy of an imaging process once deconvolution was involved?  I attach
> an old message that outlines the  argument to the end of this note.
>
> I would add to it only the fact that Hans VanderVoort ([log in to unmask], J.
> Geradtsweg 181,  1222 PS Hilversum,  The Netherlands. URL:
> http://www.svi.nl/tel: (0)35-6859405 or (0)653-345445. fax: (0)35-6837971)
> has recently had great success in deconvolving confocal data*.
>
> This software not only seems to "improve the resolution" but also reduces
> the apparent noise by taking advantage of the fact that Nyquist Sampling of
> the image of a point object (i.e. a 3D Airy Disk) with a pixel size
> suitable to recording data out to the Rayleigh Criterion (i.e. two pixels
> between the center and the first dark ring in all directions) requires
> collecting data from 64 voxels.  Knowing the PSF can effectively allow you
> to average the signal collected in these 64 voxels (or more).
>
> From now on, if one must try to compare the "resolution" of the two
> techniques, let's compare deconvolved WF with deconvolved Confocal.
>
> Jim Pawley
 
First, by resolution I mean the ability to dissociate two close objects or
the ability to extract spatial information that is consitent with
otherwise confirmed reality at a spatial frequency that is higher than the
method that one is comparing to. If one system gives me a better image,
i.e., that better agrees with reality as measured by an independent
method, then I would be inclined to call that higher resolution. The tests
that we have used would be consistent with this definition. I can see the
value of deconvolving confocal and we are working frantically on a method
to do so. I am sure that we too will find a method that works. The power
spectra of the confocal images shows that there is plenty of room for
resolution extension. I am just trying to find the best way to get that
data out. However, since our starting data for the case of a couple of
cell layers has our starting data with better signal to noise, more
linearity, and at least as high of spatial resolution, then I would not
expect that we could get even better resolution with decon/confocal for
this case. I may be wrong and I hope to be able to confirm it this summer.
In the meantime I will look further at VanderVoot's method.