Brain mask and original 3D T1-w MRI image mismatch

,
1

Hi everyone,
I work on Python 3.9.

I have to perform a brain extraction from 3D T1-w MRI images in order to obtain a whole brain mask. To do so, I used the nilearn function: compute_brain_mask.

The problem is that there is a mismatch between the computed mask and the original image when I plot them together by using the plot_roi function.

wbm
wbm938×385 118 KB

The image is a Nifti1Image object of size (152, 152, 120)
The mask and the image have the same affine transformation.

How can I fix the problem?

Thanks per advance.

2

Hi,
Can you show the cod that estimates th mask form the anat and plots it ?

(Note that mask computation in Nilearn assumes an EPI-type input. I’m not sure it works well with T1, but this is another problem than the one you’re describing).

Best,
Bertrand

3

Hi Bertrand, thanks for the help!

My code is the following:

ff
ff839×468 24.8 KB

Plot of the computed mask:

mask
mask925×320 38.9 KB

Could you suggest me a Python library to compute a brain mask of T1-w MRI?

4

Oh, I see. The function you used only applies to MNI-sampled data.
If you want to apply it to arbitrary images, you can try with compute_epi_mask(which is optimized for EPI images, but may work reasonably well for T1 images).
HTH,
Bertrand