actually you should check the talairach, which I think is why the skull stripping is failing (in fact the intensity normalization failed because of the talairach problem, which caused everything afterwards to fail) On Mon, 8 Jan 2007, Nick Schmansky wrote:

Don,

I've received the input data that you sent me which is failing the mri_ca_normalize step for you. It appears that this data fails the skull-strip step, as the brainmask.mgz is very wrong (large sections of brain are missing).

In general, it is always a good idea to check the brainmask.mgz file after running -autorecon1 (or if a failure is found when -all is run), to check that the brain appears ok (check each orientation and some slices in each, also being sure to check that the cerebellum is intact).

We have made an improvement to recon-all which addresses the skull-strip problems by running multiple instances of mri_watershed using a range of parameters, and the best skull-strip is programmatically chosen.

Attached is the recon-all to do this. Copy it to your $FREESURFER_HOME/bin and type the following:

cd $FREESURFER_HOME/bin mv recon-all recon-all.old gunzip recon-all.gz chmod a+x recon-all

Then, use the -multistrip and -clean-bm flags:

recon-all -s <subjid> -multistrip -clean-bm

The -clean-bm flag will delete the existing brainmask.mgz file, so that at the end of recon-all, the new brainmask.mgz is the one selected as the best. You would not want to include this flag if you have made manual edits to brainmask.mgz.

I have run this on your data, and it successfully creates a decent brainmask.mgz, which should allow you to run the -autorecon2 stage.

Nick

On Fri, 2007-01-05 at 11:29 -0800, Don Hagler wrote:

...

I have a subject that has gotten stuck on the CA Normalize step. I have tried running it several times and it always gets stuck in the same place, running away for days if I let it, using 99.9% of the CPU. We are running RH4-x86_64-R302.

Is this something that has been observed / dealt with before?

This is the screen output for the CA Normalize step up to the point of getting stuck: #-------------------------------------- #@# CA Normalize Fri Jan 5 11:02:19 PST 2007 /space/emc1/4/data/MMILDB/ADNI/Containers/FREESURFERRECON_005_S_0448_20060504.085355_1/mri

mri_ca_normalize -mask brainmask.mgz nu.mgz /space/monkeys/1/pubsw/packages/freesurfer/RH4-x86_64-R302/average/RB_all_2006-02-15.gca transforms/talairach.lta norm.mgz

using MR volume brainmask.mgz to mask input volume... reading 1 input volumes reading atlas from '/space/monkeys/1/pubsw/packages/freesurfer/RH4-x86_64-R302/average/RB_all_2006-02-15.gca'... setting gca type = Normal gca type reading transform from 'transforms/talairach.lta'... INFO: compute sample coordinates transform 1.405 -0.358 0.038 35.157; 0.193 0.687 -0.595 80.138; 0.176 0.895 1.062 -269.933; 0.000 0.000 0.000 1.000; INFO: transform used

---

Fixing up the home? Live Search can help http://imagine-windowslive.com/search/kits/default.aspx?kit=improve&loca...

---

Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer