[Histonet] Image analysis sample size

Bob Nienhuis bob.nienhuis <@t> gmail.com
Mon Dec 10 11:36:50 CST 2007

You may want to do a sample power calculation.
See: <http://stat.ubc.ca/~rollin/stats/ssize/index.html>
for an online version.

Bob Nienhuis
UCLA / VA Medical Center
Los Angeles

On Dec 10, 2007 7:13 AM, McGinley,John <John.McGinley <@t> colostate.edu> wrote:

> Hi,
> I have a question regarding image analysis and sample size (microscopic
> fields). Our lab has been estimating proliferation indices in IHC stained
> (BrdU and Ki-67) rat mammary gland and mammary tumors by evaluating ten 400x
> microscopic fields chosen at random, e.g. green dots placed on the tissue
> section prior to review under the scope in order to prevent selection bias
> toward heavily stained "hot spot" areas. We're using a CAS-200 imaging
> system that measures total nuclear area and DAB percent positive area above
> a specified threshold. The problem is that specimens can have a large
> range/variance depending upon the areas chosen for analysis and sampling 10
> fields may not be enough. I'm thinking the best way to compensate for this
> problem would be to increase the number of fields until the variance
> stabilizes. Does anyone have an algorithm like an Excel macro or something
> similar that addresses this problem, i.e. will estimate how many more
> fields to count beyond 10 based on current variance? Thanks in advance.
> Regards,
> John N. McGinley
> Cancer Prevention Laboratory
> Colorado State University
> john.mcginley <@t> colostate.edu
> _______________________________________________
> Histonet mailing list
> Histonet <@t> lists.utsouthwestern.edu
> http://lists.utsouthwestern.edu/mailman/listinfo/histonet

More information about the Histonet mailing list