Dear list,
I do hope this (general availability of Freesurfer) will happen soon, as the tool looks very useful indeed, and Pádraig's results shows it achieves good results without manual intervention.
You can use for a variety of important analyses that are simply not possible in e.g., FSL, SPM
*Reconstruct the cortical surface and project functional data onto it - very important in understanding topographical organization of cortical function (e.g., retinotopy) which is otherwise obscured in the volumetric data.
*Automatically identify/label cortical regions on the basis of anatomical landmarks (e.g., fusiform gyrus, motor cortex etc.) Good for anatomical ROIs
*Automatically identify/label "subcortical" regions (i.e., regions not on the cortical surface including amygdala, hippocampus, caudate etc.) Good for anatomical ROIs and volumetry
*Analyse functional group data in a more meaningful space (i.e., one in which different subjects cortical surfaces are aligned to shared sulcal landmarks) Good for statistical significance?
*Measure local cortical thickness, curvature etc. Useful for structural analysis.
I have placed some pictures onto my web page to give an idea of the segmentation possibilities http://www-users.york.ac.uk/~th512/cortical_seg_neuroimage.jpg http://www-users.york.ac.uk/~th512/subcortical_seg_neuron.jpg
Check out their publications for more information: http://surfer.nmr.mgh.harvard.edu/fswiki/ArticlesSlidesAndPosters
Many thanks to Pádraig for his sterling work in blazing this trail.
Best wishes,
Tom Hartley
(BTW I tried mailing the pictures as attachments, and but the message got stuck on the server - the mailing list doesn't forward messages over 40KB)
-----Original Message----- From: Pádraig Kitterick [mailto:p.kitterick@psych.york.ac.uk] Sent: 29 November 2007 17:34 To: ynic-users@ynic.york.ac.uk Subject: Re: Freesurfer @ YNiC - IMPORTANT UPDATE
Dear list,
Please ignore my previous instructions for installing freesurfer on YNIC machines. Gary and Mark have just pointed out a very critical point about doing this: due to the size of the freesurfer package (>3Gb), it is not feasable to have users installing their own copies of the software as the disk usage would be considerable. YNIC are looking into ways of provide freesurfer to all users from a central location sometime in the future.
Apologies for any confusion,
Padraig
Pádraig Kitterick wrote:
Dear list,
If anyone else is considering using Freesurfer (http://surfer.nmr.mgh.harvard.edu/) to reconstruct cortical surface models from T1 volumes collected at YNiC, you might be interested in the following information.
By default, if you feed a standard structural YNIC T1 into freesurfer it will most likely produce spurious brain extractions, and have great difficulty (read: 40+ hours of processing!) in reconstructing the cortical surfaces. After exploring these problems for a while, I discovered two issues which require some pre-processing to solve. The first is that freesurfer does not like images with a FOV larger than 256mm^3. Standard YNIC T1 structurals have a FOV measuring 176x290x290mm (176x256x256 slices of 1 x 1.13 x 1.13mm).
In addition to this, the voxel dimensions are misinterpreted by freesurfer when it attempts to resample the volume to 1x1x1m voxels, which it does as a standard part of the importing process. They are interpreted incorrectly as 1.13 x 1.13 x 1mm! Needless to say, this leads to all kinds of problems, resulting in spatially distored output surfaces.
Therefore, I would recommend that the following steps are taken before trying to carry out any processing with freesurfer:
- Manually resample the T1 to 1x1x1mm and force the correct
dimensions to be used, with the 'mri_convert' command (part of the
freesurfer package):
mri_convert -iis 0.9999 -ijs 1.1328 -iks 1.1328 -ois 1 -ojs 1 -oks 1 -oic 176 -ojc 290 -okc 290 T1.nii.gz T1_1mm.nii.gz
Here, we specify the input sizes (1x1.13x1.13mm), the output sizes (1x1x1mm), and I have also specified the number of output slices which is important because otherwise mri_convert will truncate them to a maximum of 256.
N.B.: This command _should_ be identical for all standard YNIC T1 structurals. However, it is always important to check that your slice counts and sizes are the same as the example given here, otherwise all subsequent processing will be compromised.
- Remove unnecessary slices from outside the head and the neck so
that the final number of slices is less than or equal to 256 in all dimensions with avwroi (from fsl):
avwroi T1_1mm T1_1mm_reslice x_start x_size y_start y_size z_start z_size
where T1_1mm is the resampled MRI (no .nii.gz extension), T1_1mm_reslice is the output volume (again, no .nii.gz extension), and the _start and _end parameters specify the starting slice and the number of slices to include in each dimension (use fslview to find the slice numbers). For example, to remove the first and last 17 slices in the Y and Z dimensions, we could run:
avwroi T1_1mm T1_1mm_reslice 1 176 18 256 18 256
which would leave us with a FOV of 176x256x256mm, compatible with freesurfer. Be careful that only redundant slices are removed!
Your pre-processed volume should now consist of less than or equal to 256 slices in each dimension, with 1x1x1mm voxels. This can be safely processed with freesurfer using:
recon-all -i T1_1mm_reslice.nii.gz -subjid <subject id> -autorecon-all
Finally, make sure you apply all of this to a copy of the T1, as you won't be able to modify the original in the mridata folder, and even if you can you really shouldn't!
I hope this will save someone a bit of time and a lot of headaches!
p.s. if you want to install freesurfer in your home folder on YNIC machines, follow these steps:
- Download the freesurfer PowerPC distribution and register for a
licence file on their website: http://surfer.nmr.mgh.harvard.edu/ 2) Double-click on the image file, and a window will open which contains a single package file. 3) Right-click on the package file and choose 'Show pacakge contents' 4) In the new window that opens, navigate into the 'Contents' folder, and drag the Archive.pax.gz to your desktop. 5) Double click on the archive file and wait for it to decompress (this will take a long time)... 6) You should now have a freesurfer folder on your desktop, which you can drag into your home folder. Now remove the archive files on your desktop as they are quite large and are no longer needed. 7) Open a text editor and copy&paste the following:
export FREESURFER_HOME=~/freesurfer source $FREESURFER_HOME/SetUpFreeSurfer.sh
and save the file as 'freesurfer-config' in your home folder. Note that this script presumes that you have moved the freesurfer folder into your home folder. 8) Copy your licence file into your freesurfer folder. 9) In an X terminal, type 'source ~/freesurfer-config'. You can now use freesurfer commands. Make sure you run this command in X11 each time you log in to have access to freesurfer.
Padraig.
-- Pádraig Kitterick Graduate Student Department of Psychology University of York Heslington York YO10 5DD UK
Tel: +44 (0) 1904 43 3170 Email: p.kitterick@psych.york.ac.uk
-- ynic-users mailing list ynic-users@ynic.york.ac.uk https://www.ynic.york.ac.uk/mailman/listinfo/ynic-users