Dear Estephan,
You should definitely not use any results where you think the segmentation is wrong.
If you think that the segmentation with fsl_anat is good, then you can just use the volume results from fsl_anat directly, rather than running SIENAX, as it does the same calculations internally to get the brain volume (normalised and un-normalised). So you should be able to skip the separate SIENAX step.
All the best,
Mark
On 14 Mar 2013, at 12:58, Estephan Moana <[log in to unmask]> wrote:
> Thanks, Mark. I noticed that some subjects did not have the grey matter correctly segmented in SIENAX. I re-ran it using the T1_biascorr.nii.gz from the fsl-anat pipeline as input, and then the segmentation looked much better. Seems like the best approach is to use both commands - fsl-anat to generate the bias-field corrected T1 then run SIENAX on it?
>
> Cheers.
>
> Estephan
>
|