2/20/96

I would like to calculate the individual areas of muscle fibers in frozen muscle sections. In the past I have used a planimeter and/or a simple grid transparency. I would like to switch to a more "1990's" method, presumably using a graphics tablet and the appropriate software. Basically, the software would not only have to calculate individual fiber areas, but also what percent of the entire muscle is made up of one fiber type or another. Certainly a built in statistical package would be useful. If anyone could recommend a set-up for use with a Macintosh, I would greatly appreciate it. If there is another way besides a graphics tablet, I would certainly consider that as well.

Thank you,
Martin A. Levin
levin@ecsuc.ctstateu.edu

The infinitely valuable (& Mac Native) image processing/analysis computer program called NIH Image written by Wayne Rasband, will readily do this work for you. You will have to first digitize an image of the muscle sections ( & a variety of options are available to you here, flat bed scanners, TV cameras, slow scan CCD's) to get the data on to the Mac, but this should be easier than using the graphics tablet. This will be your limiting step, but once done then you can digitize areas & tabulate the results. By writing a simple macro you can automatically compute % or export the results to a spreadsheet program.

Using various Operating System Emulators (Executor for the PC, MAE for Unix) the program can also be run on Unix boxes & Windoze Machines. One commerical manufacturer is developing a port of this program to the PC platform.

You can download a free copy of Image (current version number is 1.60) from either the NIH ftp site

(zippy.nimh.nih.gov)
or the MSA ftp site
(ftp.msa.microscopy.com)
or the ANL AAEM WWW Site
(www.amc.anl.gov)
in all cases you should look for the "public" directory then filter down the directory paths until you see the imaging/data processing area.

The documentation is extensive and very good.

Nestor Zaluzec
zaluzec@Microscopy.Com

Martin,
Any respectable image analysis software should be suitable. IPLab Spectrum from Signal Analytics would be good for MacIntosh. Try NIH Image first as it is free. In any case, the difficulty will be measuring the individual fibers as segmenting the individuals from the mass via thresholding may prove difficult without clearly defined boundaries. Some image enhancement might be useful. Manual tracing might be necessary.

I can supply you with IPLab spectrum. I can supply you NIH Image with a frame grabber board for your MacIntosh. SCION supplies Image with its boards.

John M. Libert
jlibert@cpcug.org

Martin,
The use of a point counting grid for measuring areas is certainly a 1990's approach and one advocated by the 'new stereology' group in Europe. Point counting is the most efficient use of time to get excellent estimates and takes less time than using thresholding or tracing of individual muscle fibers. If you combine NIH Image with some of the point counting macros on the NIH Image server you can capture video images of your tissue, overlay a point counting grid, score the hits within NIH image and export the data to excell. In general the maximum number of points that are needed for a good estimate is between 50 to 100. This combined with proper sampling of your tissue will yield quick and accurate results. See some of the more recent papers by Gunderson and group. You will find that this technique gives a more precise answer than the methods of planimetry or thresholding an image.

hagler.herb@pathology.swmed.edu
Herb Hagler, Ph.D.

I agree with the reply given by Herb Hagler, the overlay counting methods are efficient, easy to use, and belong to the 1990's battery of techniques. They can be easily used on digitized images too.

However, I would like to add an additional point. If you are interested in individual fiber areas then working with the selected fibers is OK. If you are more interested in the mean fiber volume then you must take care to properly sample the whole population. Working on a few selected pictures or tracings will add a bias to your result and thus disqualify it.

To properly eliminate bias requires a defined sampling strategy which starts at the animal (or culture dish) level and continues through embedding and sectioning to photography. One paper to cover this point in a readable way is in J. Histochem. Cytochem. 40:1929-1936 1992 and is written by John Lucocq.

Paul Webster, Ph.D.
paul.webster@yale.edu

[Return to Tips & Tricks Menu]