Hi Allison,

you have two options:

1. replace the mri_robust_register and mri_robust_template executables with newer ones (but who knows if flags changed, so this may not work directly, in that case also the call in recon-all may need to be adjusted).

2. better: simply run all bases with 5.3 and also all the longitudinal runs with 5.3. I think this should work based on 5.1 cross sectional data (you may want to test in a single case). As long as you do it for all subjects this will be ok.

Best, Martin

On Dec 8, 2016, at 3:51 PM, Allison Rainford allison.rainford@northwestern.edu wrote:

Hi FreeSurfer team,

My lab is looking at longitudinal data across 5 time points for each subject. All of our time points have been edited in version 5.1 for consistency. However, I have run into an issue when trying to create the bases for these subjects in this version due to a memory allocation problem. It appears that 5.1 cannot allocate enough memory to store the temporary files for each of these time points and consequently errors out during the registration stage with this message:

Iteration(f): 2terminate called after throwing an instance of 'std::bad_alloc' what(): St9bad_alloc

I also tried creating the base using a memory flag (-m) to allocate more memory but had no luck in those attempts.

I tried creating the base in version 5.3 and this ran successfully. However, because this is a longitudinal data set ran in version 5.1, it’s problematic that the jobs crash in 5.1.

Is there anyway to allocate sufficient memory in 5.1?

Thank you,

Allison Rainford Research Study Assistant Cognitive Neurology and Alzheimer’s Disease Center Northwestern University 320 East Superior Street Searle 11-579 Chicago, Illinois 60611 Email: allison.rainford@northwestern.edu

---

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