How can I segment brain gray matter from white matter?

Hi. How can I segment the gray matter from the white matter from the attached image? Please help me with MATLAB codes.

[MarKf]

Your jpeg seems to be a black and white image, you'd need at least a grayscale one to segment gray and white matter (also a neuroscientist may understand what you mean with gm and wm, but this question -same as the pic- is awfully generic).

You can then use a threshold to separate them in a single image. if the image has a gradient or a local bias a local adaptive thresholding method might be useful. There are tools to get the threshold automatically and to create probabilistic maps of the tissues. One example is ft_volumesegment from the fieldtrip toolbox.

[Prince Lassemento]

Ok! here is another image. Please help me with a matlab code to segment gray matter from matter.

Answer 1

[Image Analyst]

Find published algorithms here:

http://www.visionbib.com/bibliography/contentsmedical.html#Medical%20Applications,%20CAT,%20MRI,%20Ultrasound,%20Heart%20Models,%20Brain%20Models

Pick one and code it up yourself, or ask the author to buy the code from them.

[Image Analyst]

See my Image Segmentation Tutorial in my File Exchange:

Image Analyst's File Exchange

It's a generic, general purpose demo of how to threshold an image to find blobs, and then measure things about the blobs, and extract certain blobs based on their areas or diameters.

[Prince Lassemento]

Thank you. But I am NOT understanding the interpretation of your codes because working with matlab and coding in matlab is new to me. Can I schedule a zoom meeting with you? [e-mail] is my email. I am a first year M.S Research Scholar at IITM, in the department of Biomedical Engineering.

[Image Analyst]

Sorry, I don't do private consulting for Answers people.

To learn fundamental concepts, invest 2 hours of your time here:

MATLAB Academy - Free 2 hour training

Answer 2

[Walter Roberson]

First of all: you have zero possibility of accurately segmenting a JPEG image (unless it is one of the rare lossless JPEG images). Lossless JPEG blurs edges, and you will not be able to recover from that.

Secondly: A place I used to work spent several years trying to create algorithms to segment brain MRI images, working with brain surgeons and neurophysicts. 72% accuracy we could do routinely; to get past 78% accuracy required a lot of work, at least with real brain MRI.

What we found, in practice, was that getting useful accuracy required switching out of using MRI or CT images, and to instead use MRS -- to effectivelyu do chemical analysis voxel by voxel instead of doing image analysis.

Perhaps our results would have been different if we had had access to Recurrent Neural Network (RNN) technology. Our shallow neural network work did not produce any useful results -- too many false positives, too many false negatives.