Hey Allison,
I appreciate you getting back to me. Fortunately I figured out the problem. We have some strange issue with one of our lab computers (it's running Red Hat but we had problems with the FS rh9 distro so we were using CentOS version of FS), and basically it did not show the updated segmentation that had occurred after running autorecon2 (after adding control points). Instead it only showed the control points and the old segmentation. Pretty weird, but there's definitely issues with that computer because tksurfer does not work either. A grad student we work with noticed this discrepancy when he viewed the data on his own machine. So, I was fairly confident in what I was doing but totally confused with what I was seeing, hehe. Right now just trying to get better segmentations from sagittally acquired data and will be looking into FS-Fast for functional analysis -- previously we've used afni. Our segmentations have tended to work out much better with axially-acquired images, but we're getting there..
Anyway, thanks!
John
---------- Forwarded message ----------
From: Allison Stevens <astevens@nmr.mgh.harvard.edu>
Date: Oct 1, 2008 12:39 PM
Subject: Re: [Freesurfer] Fixing WM intensity normalization via control points
To: John Sheppard <johnsheppard2007@u.northwestern.edu>
Did anyone answer these questions for you?
On Wed, 24 Sep 2008, John Sheppard wrote:
Hello,
I have some basic questions with regards to adding control points to fix the
WM intensity normalization. I am running some subjects through autorecon 1
and 2 only (we are not concerned with subcortical structures). The
reconstructed volumes had a lot of missing grey matter in the lateral
temporal lobe, and my intention is to fix this by adding WM control points
(the intensity is also off for WM in places). Long story short, I have the
control points added and saved, and then I run the subjects through
recon-all autorecon2-cp.
1) Am I correct that autorecon3 will not be necessary if I am only including
cortical structures? (I see it is included in the command line on the
control point tutorial.)
2) Once I run the subjects through autorecon2-cp, when I open them in
tkmedit the segmentation still has not changed - the control points are
still there as before. Do I need to run the subjects through some other
processing step before I will see the changes? (I am doing tkmdedit SUBJECT
brainmask.mgz lh.white -aux T1.mgz -aux-surface rh.white)
3) Based on the workflow in
http://surfer.nmr.mgh.harvard.edu/fswiki/RecommendedReconstruction, it seems
that control points need to be added directly after autorecon1. I am
actually adding the CPs on subjects that have gone through autorecon2
already. Is this a problem or is it still possibe? (If a problem I can
reprocess the data.)
Sorry if these are painfully obvious questions, but I seem to be slightly
confused with the processing pipeline.
Much thanks,
John Sheppard