Using Image->Adjust->Threshold, you can achieve the same result as
pasting the histogram into Excel and choosing a cutoff. Note that
threshold only works on 8-bit grayscale images, and when thresholding
the histogram displayed in the dialog is for the ROI if one is
selected (otherwise it's the histogram of the whole image).
In Analyze->Set Scale, you can set a conversion factor to change
pixels to some real world unit. You will need a scale mark in the
image or some knowledge of your imaging setup; a scale mark is the
easiest solution. In Analyze->Set Measurements you can choose the
results you want to measure. You should have "Limit to Threshold"
selected, and when you choose your threshold, make sure that the
portion you want to measure is red. Muscle tissue in your case.
When you have set the scale and the measurements you want, you can
select your ROI, pick the threshold, then use Analyze->Measure, and
the results will be shown in the results table. If you want to have
multiple ROI, such as in the Rapidia image you linked, you can try the
ROI manager (Analyze->Tools->ROI manager).
ImageJ will list the results in the results table (which can be
saved), but if you wanted you could make a macro to draw them onto the
image like Rapidia does. I like my results in text files though since
they are easier to deal with later.
That should speed things up a bit. If the scale doesn't change
between images, you can leave it alone and process several images at a
time. As far as more automation, if your images are like the one you
linked, then you will almost certainly need to draw the ROI by hand.
However, you may be able to use an autothreshold instead of choosing
one manually. ImageJ's threshold dialog has an auto button that might
work, but if you check the plug-in site, there are thresholders that
use different algorithms. Also you can read about the macro language
(
http://rsbweb.nih.gov/ij/developer/macro/macros.html) which can speed
up repetitive tasks, like naming ROI.
Justin
On Sat, Dec 13, 2008 at 1:22 PM, casab <
[hidden email]> wrote:
> On MRI images, I am trying to measure the cross sectional area of muscle
> after subtracting out the fat component. I figured out one way to do it
> with ImageJ, but it is very labor intensive. Basically, I draw a region of
> interest, make a histogram of the pixel intensities, cut/paste the histogram
> list into excel, manually choose a pixel intensity cut-off between fat and
> muscle, count pixels on "muscle side" of cut-off, and then convert the
> number of pixels into an area measurement. Does anyone know of a faster or
> automated way to do this? Perhaps there is a different software program
> that is free or inexpensive that can do this? The attached image is an
> example of what I would like to do, performed with a Korean program called
> Rapidia. Unfortunately, that program is quite expensive. Any ideas?
> Thanks
>
>
>
http://n2.nabble.com/file/n1652546/multifidus.jpg>
> --
> View this message in context:
http://n2.nabble.com/Cross-section-area-of-muscle%2C-less-the-fat-tp1652546p1652546.html> Sent from the ImageJ mailing list archive at Nabble.com.
>
Locked
