Hi Amanda,
there can always be individual outliers (in both the longitudinal and even more so in the cross sectional stream). These can be causes by low image quality or motion artifacts etc, or by failure of some steps in the processing stream. That is why we recommend to do QC (e.g. check skull strip. talairach transform, wm , surfaces etc). You should open up these cases and take a close look at what is going on.
On average (if you have sufficiently many cases in you set, like 15 or more) you should see that the across time variance is reduced when using the longitudinal stream.
Best, Martin
On 11/12/2015 11:29 AM, Worker, Amanda wrote:
Hi All,
I have processed some longitudinal data (0, 1 week, 4 weeks) using the advised longitudinal stream. This is a healthy control sample and I wasn't expecting to see any differences in cortical thickness over the time points.
Looking at some of the measures I have noticed that something doesn't seem quite right
e.g. for one subject the insula measures:
Cross-sectional Longitudinal Base
A 2.835 3.016
B 2.721 1.864
C 3.050 3.107
3.098
The majority the longitudinally processed data looks reasonable, but there are a few instances where one timepoint looks completely different.
Do you have any ideas if something may have gone wrong? I have attached the recon-all.log files for the template and the 3 longitudinally processed timepoints.
Thanks in advance for your help,
Amanda
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer