Fwd: ImageJ Analyze Particles with Varying Magnification Issues

[Suzie Shdo]

I am analyzing a series of SEM images of quartz particles using ImageJ to
measure particle size distribution. I observed that ImageJ calculated
different average particle sizes for the same particles at different
magnifications.

Generally, the higher the magnification the smaller ImageJ estimated the
average size (for the exact same particle). I converted these values to
particle diameter (assuming circular) and observed that the diameters did
not seem accurate as compared with the scale bars generated by the SEM. To
confirm this discrepancy, I cropped an individual particle out of two images
at two different magnifications and confirmed that ImageJ was giving me
different values for the exact same particle.

I can send an image at 50,000X magnification in which I believe ImageJ is
underestimating the size of the particles to give you more information.

To analyze this image, I am setting the scale based on the width of the
image and the number of pixels. Then I am manually adjusting the threshold
and using the find edges tool to outline the particles. Finally, I am using
the analyze particle tool to collect the average particle size and standard
deviation. I use this average size value to convert to particle diameter
(assuming circular particles).

Please let me know if you could offer any advice about this issue.

Thank you for your time.

Sincerely,

Suzie Shdo

*Research Assistant
Environmental Health
Harvard School of Public Health*

[Gabriel Landini]

On Tuesday 08 Feb 2011 17:36:56 you wrote:
> I am analyzing a series of SEM images of quartz particles using ImageJ to
> measure particle size distribution. I observed that ImageJ calculated
> different average particle sizes for the same particles at different
> magnifications.

Most SEMs insert a calibration based on some setting of the microscope, and
not from an object of known size. Those are not particularly accurate.

This should be possible to calibrate, but you might have to call the technical
service (i have seen somebody calibrating the instrument using TEM copper
grids, which you could measure with a calibrated optical microscope).
I would suggest to insert an object of known size and image it at various
resolutions to see if the calibration reported is true.
If not and assuming there is a maintenance contract, you could ask them to do
it for you.

Cheers

G