[awenlou29]

Hi,
When I want to perform the Skull Stripping with my MRI images, it results with a terrible mask, not even surrounding the brain. I tried changing all the parameters, but the same problem is still there. Laso I have the “maximum value is 0” messages with these images. Is it due to the images in themselves ? How can I go through this problem ?

[Anand Joshi]

It is possible that there is a header issue in your images. Please check if the header in your images is correct. Based on the error message you mention it seems that the values are all 0 or negative.
To check the header.
Please 1) open the image in BrainSuite,
2) Goto Tools->Volume Info

If you still have the problem, if you can post that information here I can have a look.

[awenlou29]

Hi,
I did it, and here is what I get for one of the images:

176*256*256
1mm*1mm*1mm
data type: sint16
description:
position: -87.5 -119.715 -140.245
file orientation: identity matrix (3×3)
working orientation : identity matrix (3×3)

I compared with one of the images provided by Brainsuite for training, and the only differences are the data type and the file orientation (the first term is -1 in the provided example’s matrix instead of 1). Could the problem come from here? Is it fixable?

[Anand Joshi]

To me, the header info looks good and should not be a problem. Is there a long neck in the acquisition? The only time when I see an issue with the skull stripping is when there is a long neck or if there is some imaging artifact. You can use the mask tool in BrainSuite to clean up the images or remove the neck. If that doesn’t work if you can share the image or screenshot I can have a look.

[awenlou29]

Hi,

I managed to fix the problem. I used an other visualizer (ITK Snap) to open the image, then saved it without touching anything in the same format. Then, the BSE worked perfectly with the “new version” of the image in Brainsuite. I don’t really understand why it worked, but it worked!

Anyway, thanks for the help and reactivity!

[Anand Joshi]

Great! it probably is due to some header issue in the original image. Glad that it is working now.

[Author]

Posts