Search the CONFOCAL archive at
http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal

Hi,

Considering the CV of measurements obtained from (confocal) microscopy
systems. Sampling a biological structure, e.g. a nucleus, at a depth
which is less than its thickness in a sample in which their Z-position
also varies, will of course deteriorate the Cv of the quantification.

When imaging a sample with a scanning/confocal microscope, you make
Z-slices through a samaple which is a 3D structure and of which all
objects, e.g. nuclei, are not in the same Z-plane. The resulting image
will indeed look beatiful, but if you do you measurements on one layer
of the Z-stack, e.g. area, you will of course see considerable variation
in your measurments as the nuclei will be sliced at different positions
of a "sphere". Most of the scanning cytometers and scanning image
systems on the market create nice images, but do not take the
consequence of Z-sampling in their quantification software.

When measuring the diameter of a nucleus, ths software should detect the
3D centre of gravity of the structure and then measure its diameter.
When doing densitometry, the samples taken should also be matched in
respect to location and type of structure or should do the calculations
on the entire 3D structure.

Using a widefield microscope with a wider depth of field (DOF), captures
less of this variability to start with and as such reduces the
Z-variation efect in this respect.

For digital microscopy and measurements on digital images the following
artice is ver nice to understand the sampling issue in digital microscopy::

Quantitative Microscopy by Ian T. Young of the T.U. Delft

Webpage http://www.ph.tn.tudelft.nl/People/young/manuscripts/QM/QM.html

A large C.V. coming out of a digital microscope is mainly caused by
spatial undersampling.

Sampling a the Nyquist rate gives you a nice picture, but not yet good
quality measurements
(http://ourworld.compuserve.com/homepages/pvosta/pcrnyq.htm)

With appropriate spatial sampling, the CV of quantitification due to the
instrument can be brought down to about 1 percent, but the sytem will
still capture the biological/experimental variation, which is often more
than 10 percent. Using a calibrated sample is the best way to measure
the CV of quantification coming out of a digital microsocpy system at a
given resolution ( ~N.A.) and magnification.

There is a lot to to say about quantification in digital microscopy and
  flow cytometry, but this would require an article on itself.

Regards,

Peter Van Osta

----------------------------------------------
Peter Van Osta

Director Imaging
MAIA SCIENTIFIC (formerly Union Biometrica NV)
Cipalstraat 3
B-2440 Geel, Belgium
Tel.: +32 (0)14 570 620
Mobile: +32 (0)497 228 725
Fax.: +32 (0)14 570 621
Email: [log in to unmask]
Website: www.maia-scientific.com
A Harvard Bioscience Company
----------------------------------------------

=================================================


>>"In flow cytometry we can measure DNA content or beads with high
>>precision - 0.9 - 1.5% DNA CVs reported.  In one
>>report, with 1.5% CVs, the machine electronic noise was measured at
>>~0.5% CV.  I have
>>never seen imaging systems that can do this.  A while back I obtained
>>some laser scanning
>>microsope images of DAPI stained Hela, and the CVs were terrible
>>(40 -
>>50%, maybe).  These same
>>cells yield DNA histograms with 4-5% CVs on flow cytometers,
>>routinely.If we look at DNA or beads on laser or lamp scanning
>>*cytometers*(iCyte, Q3DM) the CVs are not as good as flow
>>cytometers, but are
>>considerably better than any data I've seen (rarely) from a system
>>thatwas designed with imaging in mind.  Therefore, I would like to
>>know (1)
>>are there inherently physical/engineering reasons why slide-based
>>systems are less precise than flow (liquid) -based systems, and
>>(2) is
>>the primary reason that laser scanning microscopes are not as
>>quantitatively precise as laser scanning cytometers because the
>>engineers are image (feature/morphology) oriented rather than
>>quantitatively oriented, and the optimizations are so ordered?, or (3)
>>am I ignorant on this issue (i.e., there are good quantitative data
>>generated and published via laser scanning microscopes)?  I am writing
>>an editorial, and I would like to summarize the differences
>>between flow
>>cytometers, scanning cytometers, and scanning image systems in
>>terms of
>>quantification."
>>
>>Best wishes
>>
>>Bob