I don't know if this will be helpful, but we had a similar issue, which was addressed by Mark last year. In short, this was the pipeline that seemed to correct it most of the time:
- run brain extraction on one of your functional images (e.g. example_func)
- dilate the brain mask using fslmaths and -dilF that you get from this (since there might be motion between the functional scans, so we don't want the mask to be too tight and exclude any actual brain tissue in other scans)
- apply the mask to the whole 4D functional dataset (with fslmaths and -mas)
- use this masked version as the input 4D dataset for FEAT
Worth a shot!
|