Print

Print
Hi Jesper,

I've uploaded "jesper.tar.gz" to FSL_Uploads on OxFile.

$ tar -cvzf jesper.tar.gz ./jesper/*.nii.gz
a ./jesper/1_T1w_raw.nii.gz  #this is our raw T1 image
a ./jesper/2_T1w_acpc_brain.nii.gz  #this is T1w_raw rigidly aligned to the axes of MNI space (i.e., acpc) and it is our FNIRT input
a ./jesper/3_T1w_fnirt_brain.nii.gz  #this is the FNIRT output

Thanks for taking a look at it.

Simon


On 25 January 2016 at 20:40, Jesper Andersson <[log in to unmask]> wrote:
Hi there Simon,

it is hard to say without seeing your data. One known problem is when your images have a high signal from marrow in the cranial bone compared to what is present in the MNI template. If you tar and upload your T1 image before and after fnirting I can have a look at it and perhaps give some advice.

Upload link: https://oxfile.ox.ac.uk/oxfile/work/extBox?id=72139C7E463068D5F 

Jesper


On 25 Jan 2016, at 04:11, Simon Baker <[log in to unmask]> wrote:

Hi all,

We used HCP's PreFreeSurferPipeline.sh to non-linearly register the T1w volumes in our dataset to the MNI152 template brain (MNI152_T1_1mm_brain.nii.gz) and found that the output brains appear larger than perhaps they should compared to the MNI152 template brain. For example, X,Y,Z coordinate 91,183,41 is located at the perimeter of the output brains but the same coordinate is located well outside the perimeter of the MNI152 template brain. How can we optimise FNIRT to get a more accurate non-linear registration?

Thank you,

Simon Baker
Brain & Mental Health Laboratory
Institute of Cognitive & Clinical Neuroscience
Monash University