RE: [Histonet] quantitative graded scale for IHC
I would agree with the comments below from Greg.
The major problem with comparing stain intensities between sections is
that you generally have no real measure of the section thickness for an
individual section. If you are cutting at 5 microns there can easily be
a one micron difference between sections, which translated into
differences in volume of tissue and intensity of staining can be
significant. One way around this is to prepare a block of tissue that is
embedded with the sample. The block can be of a homogeneous protein
material stained for example with a Procion dye.
You can, if you have no social life, prepare sections of this material,
dissect out a measured area with a blade (using a dissecting scope) and
weigh it. You can then relate this volume of tissue to its intensity of
staining and work back to relate this to section thickness.
If a cylinder of this material is embedded with the block it provides a
built in control that will allow comparison between sections.
his does not of course take into account any differences in staining due
to differences in staining conditions for different slides.
[mailto:email@example.com] On Behalf Of Greg
Sent: Thursday, September 09, 2004 7:54 AM
To: Carla M Conway
Subject: Re: [Histonet] quantitative graded scale for IHC
I certainly may be corrected by others on this but I personally don't
think you will obtain statistically valid data by grading intensity of
reaction between cells or fields or sections. There is just so much
room for subjectivity and for that matter, variability between runs or
perhaps even between slides within runs!
What you may want to try is calculating a ratio of positive to
negative (perhaps pos. cells vs. neg. cells in a given area). For
instance: place a grid with 50 or 100 intersects on the computer
monitor, choose a microscopic field (in a random or systematic-
random manner), and then at each intersect decide whether the
reaction at that point is positive or negative. If at any particular
intersect you have trouble to decide whether it is pos or neg, have
a system that does not permit subjectivity, such as: for such points
where the call could go either way, look at the space imediately to
the left of the y-axis and above the x-axis and then make your
decision. Determination of the number of fields you need to sample
in order to obtain statistically valid data is very important and a
matter for a statistician to explain and/or determine for you (which I
most definately am not) unless you happen to have that training
Another idea would be to have the image analysis software either
measure the area of or count the number of pixels in a given area
(field) that are "this red or redder". This involves choosing a
relatively arbitrational threshold for the degree of redness that is to
be measured or counted and applying the exact same setting to all
other fields and sections in the study. My personal experience
(using Bioquant NOVA) has been this sounds great in theory but
was very difficult to pull off! The good old-fashioned counting was
much more reliable and for that matter, easier to defend in
presentation or publication.
I am eager to hear other ideas or methods that might also work!
From: "Carla M Conway"
Date sent: Thu, 9 Sep 2004 07:22:02 -0700
Copies to: Subject: [Histonet] quantitative
graded scale for IHC
> I would like to devise a quantitative graded scale for IHC results
> would provide more info than just +/- staining.
> I have a reference where IHC staining intensity was rated (from pale
> to dark red) using alkaline phosphatase/Vector-Red, however we are
> Envision+/AEC which seems to be an all or nothing (red or no red)
> with no gradation. We just received ImagePro image analysis software,
> this may be another route to try. Thanks in advance for any comments
> suggestions you may have.
> Carla Conway
> Western Fisheries Research Center
> Seattle, WA
> Histonet mailing list
Atlantic Veterinary College, U.P.E.I.
550 University Ave.
Canada, C1A 4P3
Happiness is a journey, not a destination.
Histonet mailing list
Histonet mailing list
<< Previous Message | Next Message >>