Stephen,

The 30 minute estimate for the -autorecon1 stage is no longer accurate. It got slower when we added a feature to improved the skull-strip step. The feature is to make use of an atlas to determine where skull and cerebellum might be. To use that atlas requires a registration to that atlas, and it is that registration (mri_em_register), that adds the extra 60 minutes.

However, we are looking into using the talairach.xfm transform in place of the talairach_with_skull.lta transform generated by mri_em_register for this atlas registration. So possibly the next release (v4.3.0) would return to the 30 minute -autorecon1 stage.

Nick

On Wed, 2009-03-04 at 21:47 -0500, Stephen Towler wrote:

Hi all. Documentation estimates 30 minutes for -autorecon1 on the sample data provided with the most recent freesurfer release. Our 2008 iMac with 4GB of RAM takes about 90 minutes. Throughout the job CPU utilization was high but there was plenty of free RAM.

Is 90 minutes common for similarly equipped leopard systems?

Thanks, Stephen

--

Stephen Towler Graduate Student, Neuropsychology Dept. Clinical and Health Psychology

stowler@phhp.ufl.edu calendar: http://cli.gs/9nvvDL 352-273-8715 lab (Shands G078) 352-273-6015 lab (HPNP) 352-682-5231 mobile 352-273-6156 fax

Price Lab PO Box 100165 101 S. Newell Drive University of Florida Gainesville, FL 32610 _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer