Dear Gillian,
The brain extraction is done differently with fsl_anat - it uses a registration-based method instead of the standard BET processing. The --betfparam is used to adjust the initial, rough brain extraction that is done as part of getting the registration to work. So this option is really only useful for debugging early stages of the pipeline. If you have small errors in the brain extraction then this will not help. If that is the case you can either use fslmaths to erode or dilate the mask, or revert to using BET instead of fsl_anat. If you are not satisfied with the fsl_anat results, I'd like to know what is wrong with them in order to help improve its performance.
All the best,
Mark
On 5 Dec 2012, at 17:43, Gillian Cooke <[log in to unmask]> wrote:
> Hi Mark,
>
> Thanks for that suggestion. The command now works; however, I don't see any change when I run the BET at different parameters. I am correct to look at the T1_biascorr_brain.nii.gz right? This is the output file that we should use?
>
> I also see no differences when I look at the different T1_biascorr_brain_mask.nii.gz files.
> When I used my old scripts I can see a difference between the _brain and _brain_mask files at different BET levels.
>
> Any suggestions as to what might be going wrong?
>
> All help gratefully appreciated :)
>
> Gillian
>
|