Generating seperate histogram curves for different brain tissue in T1 and T2 MR Images

Dear all,

I am working on brain MRI segmentation. In some studies I have seen that they could create separate curves in histogram for different parts of the brain, like "White matter", or "Gray matter", and CSF (CerebroSpinal Fluid) for t1 and t2 MR images. An example is shown in bellow.

The left-hand image is T1-weighted, the one on the right is T2-weighted.

And these are the created histograms:

I could create histogram, for the whole brain, but not separately for different tissues/parts.

Now, my question is:

How is the process?, did they first extracted different parts then created histogram for all parts and showed them all in one figure, or is it possible to create the separate histogram curves for different tissues all at once.

Any Idea how to do that?

thanks.

[Nasrin Akter]

Hello

Did you generate these T1 and T2 weighted images? Is there any code available?

Answer 1

[Image Analyst]

First you have to segment the images. From the looks of the histograms I can tell that, because they overlap, that you cannot simply use gray level range to do the segmentation. You'll have to use additional other criteria, such as shape or location. Once there's segmented you'll have a mask for every region and you can use logical indexing and histogram or histcounts to make the histogram

histogram(grayImage(gmMask));
histogram(grayImage(wmMask));
histogram(grayImage(csfMask));

For algorithms to get the various tissue type masks, see http://www.visionbib.com/bibliography/contents.html

[Cyrus]

Thank you for your response, any idea for segmentation algorithm? our goal is to segment the image into three clusters, WM, GM and CSF.

Best.

[Image Analyst]

My idea was the last sentence of my answer.