Hi Myriam
we are continually working to improve accuracy and robustness. Amygdala is of course pretty difficult. Did you look at the discrepant case? If so, which one was visually more accurate?
And yes, we expect that somewhat less editing is required in 5.1. There is also an array of new functionality in 5.1 that wasn't in 5.0 (e.g. hippocampal field segmentation, automated tractography, combined volume/surface segmentation, improved longitudinal, etc....)
cheers Bruce
On Mon, 11 Jul 2011, Myriam Siefert wrote:
Hello Freesurfer users and experts,
We already run 55 subjects with Freesurfer 5.0 and wish now to increase our sample size up to 200 and are now deciding whether to run everybody again with Freesurfer 5.1 or just go on with the previous version. We are doing subcortical analysis (amygdala) and hopefully cortical analysis (correlations with behavioural results), on Siemens 3T.
I have 2 questions relative to this issue :
- To what extend should the results be more accurate or reliable (I read in the release note that 3T data is now better handled) ? I run 3 subjects on boths versions and compared aseg stats, most results were very close except for a few that were highly discrepant (ex left amygdala and hippocampus for 1 subject were much smaller with FS 5.1). What could explain this?
- I also noticed on the web site that the recommended workflow had been updated. Is this linked to the new release (if less or no editing is needed, checking the output
only after recon-all -all would obviously save time...)? Is less need for editing expected with the new release? Or is less editing now advised ("Less is more")?
Thanks in advance for any advice!
Myriam Siefert