External Email - Use Caution
Dear Freesurfer team,
Thank you for your continued efforts and advances on the Freesurfer tool, which have benefitted the community and quest for therapy advances. I wondered whether there is an assessment of the computation time breakout between volume calculation vs. cortical thickness, area, and mean curvature. Is it possible to reduce time by calculating only volumes (no CT, area, MC), and if so, what reduction would be anticipated (recognizing that the answer is proportional as the absolute time would depend upon hardware etc.)? As a second question, there was a previous mention in listserv comments regarding running left and right hemispheres in parallel. Is this done, and is it an approach that can be combined with the parallel processing flag? Thank you for any insights on these questions.
Dawn
External Email - Use Caution
Hi Dawn,
See https://surfer.nmr.mgh.harvard.edu/fswiki/recon-all#Step-wiseDirectives-1 for an approximate beakdown of computational time per step. The parallel flag in FS v6 will decrease this to about 3 hours total on a 4 core machine.
(time varies depending on your workstation, surface defects, etc; if using a cluster, check with network admins about using -parallel)
Volume, thickness, area, curv are all going to be seconds to minutes calculations and would not meaningfully impact your recon-all time.
-Adam
On Fri, Nov 16, 2018 at 3:47 PM Dawn Matthews dmatthews@admdx.com wrote:
External Email - Use CautionDear Freesurfer team,
Thank you for your continued efforts and advances on the Freesurfer tool, which have benefitted the community and quest for therapy advances. I wondered whether there is an assessment of the computation time breakout between volume calculation vs. cortical thickness, area, and mean curvature. Is it possible to reduce time by calculating only volumes (no CT, area, MC), and if so, what reduction would be anticipated (recognizing that the answer is proportional as the absolute time would depend upon hardware etc.)? As a second question, there was a previous mention in listserv comments regarding running left and right hemispheres in parallel. Is this done, and is it an approach that can be combined with the parallel processing flag? Thank you for any insights on these questions.
Dawn
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu