That might be quite challenging as BET was probably designed for MRI
images.  Another way you can brain extract images is to register them
nonlinearly to a target where you have already defined a brainmask and
then invert the transform and bring the mask back to the individual.

Peace,

Matt.

On 2/15/17, 7:57 PM, "FSL - FMRIB's Software Library on behalf of Sam

Choi" <[log in to unmask] on behalf of [log in to unmask]> wrote:

>Dear FSLers,
>
>I am working with CT scans of stroke patients and would like to use FSL's
>brain extraction tool (BET) to remove the skull from the brain images.
>
>When I use the BET command (below), the resulting brain image still
>includes the skull and other artifacts.
>bet ct_scan.nii.gz ct_scan_brain.nii.gz
>
>I looked on the FSL forum but did not find much information on CT scans.
>I am wondering if I should experiment with the -f option (fractional
>intensity threshold, between 0-->1) in the BET command until I find an
>appropriate brain outline estimate. For example:
>bet ct_scan.nii.gz ct_scan_brain.nii.gz -f 0.9
>
>Given that I am working with CT scans of many patients, it is likely the
>-f value will vary for each brain that I am skull stripping, which makes
>this solution quite subjective. I am wondering if there is a more
>'systematic' approach to carry out BET for CT scans (e.g., change voxel
>intensities of the CT scans to create better contrast between skull and
>brain).
>
>Thanks for your help!
>
>Sam