Hi Yoonho

It mostly applies to the nonlinear warp as the “further” an anatomy it is from our atlas the longer it will take to compute. And yes, that is reasonable definition of cnr (although you would need some squares/sqrts or abs). The higher the CNR the better the segmentation and the fewer the topological defects.
Acquisition optimization is way more important as it impacts accuracy in addition to time.
It is certainly possible but might not be wise. Can you tell us more about this? Are all your time 1s on one scanner and all time 2s on another? Is there a distribution of subject types/groups or some other demographic variable you are interested in?

Cheers

Bruce

From: freesurfer-bounces@nmr.mgh.harvard.edu <freesurfer-bounces@nmr.mgh.harvard.edu> On Behalf Of Yoonho Hwang
Sent: Thursday, September 8, 2022 6:40 AM
To: freesurfer@nmr.mgh.harvard.edu
Subject: [Freesurfer] Shortening recon-all process time and longitudinal process

External Email - Use Caution

Dear Bruce Fischl,

Hi, I'm Yoonho.

In the previous post, "Re:[Freesurfer] Inquiry about process time by use "recon-all" (Fri, 12 Jan 2018 12:29:19-0800)", I found that the process time for recon-all depend on both how abnormal the anatomy is and how high the CNR of your scan is.

Does "how abnormal the anatomy" mean that there is an anatomical abnormality due to a disease based on MR image and "how high the CNR of your scan is" mean that there is a CNR, [x_mean (gray matter) - y_mean (white matter)] / standard deviation_air (background)?
Related to process time for recon-all based on the FS ver.7.2.0, is open-mp effect stronger than acquisition parameter optimization effect? Which one is more helpful in reducing the process time?

In the longitudinal analysis for freesurfer,

Is it possible to use longitudinal analysis of MR images obtained from equipment with different magnetic field strengths (1.5T vs 3T) in time series for the same subjects?

Looking forward to your feedback.

Thank you very much.

Sincerely,

Yoonho