Richard-
 
We have not seen any of those problems with our system. The Sedat/Agard
reiiterative constrained algorithm goes to great lengths to reduce
artifacts caused by the Fourier math. Of course, an image is only as good
as the original and the microscopy. This is why we currently prefer the
whole system solution that API is selling. The algorithm is only one part
of the whole process. In fact, our starting images on the deconvolution
system are often as good or better in resolution (and certainly better
signal-to-noise (s/n)) than our CSLM images. We have done the experiment of
grossly under exposing an image to see if artifactual "data" appears. All
that is seen is a systematic reduction in s/n.
 
We have recently hashed alot of the comparisons between confocal and
decon. Unless people want an update of our findings, I would suggest that
if you want more details that you contact me privately and I will send
you more details on what we have found. This way we will spare the masses
if they are not interested.
 
________________________________________________________________________________
 
 
Paul Goodwin
Image Analysis Lab
FHCRC, Seattle, WA
 
On Fri, 1 Sep 1995, Richard Thrift wrote:
 
>             -Reply)
>
> Thanks, Paul, that's very helpful.
> Do you or other respondents have any comments on other limitations of
> deconvolution (besides processing time, and use for thick samples)?
> Are there artifacts that show up under certain circumstances?  Problems
> with resolution, intensity, etc.?
>
> Thanks
> Richard Thrift