On Thu, Aug 20, 2009 at 12:02 AM, Mark Jenkinson
<[log in to unmask]> wrote:
Hi,
There have been some quite significant changes made to FIRST
since it was used for this paper. Therefore the comparisons
are not up-to-date anymore. What we have found is that FIRST
can have trouble on severely atrophied hippocampi, as it does not
contain much training data for this. We are currently working on
ways of improving this, so can potentially help if you have some
severe atrophy in your subjects. If not, then I think FIRST is
working reasonably well with the hippocampus, and better than
was reported by Morey et al.
I'm quite surprised that you are also getting good results on T2
images, but am pleased to hear it!
All the best,
Mark
On 20 Aug 2009, at 04:10, Dianne Patterson wrote:
Dear Group,
I've been reading
A comparison of automated segmentation and manual tracing for quantifying hippocampal and amygdala volumes
by Rajendra A. Morey et al. NeuroImage 45 (2009) 855–866
They compare First and Freesurfer and conclude that First is not as close to their expert manual segmentations of the hippocampus as is Freesurfer. However, they were using an earlier version of FIRST [FIRST (FSL v4.0.1)]
Have specific improvements have been made to FIRST hippocampus segmentations since that time? I certainly can't match their manual expertise, so I'd like to use FIRST as a jumping off point.
By the way, I'm pretty impressed with what FIRST can do automatically, I know it is intended for T1s, but it did a pretty good job on my partial (not whole-brain) T2 images...even aligning a lot of them really well....and often finding the hippocampus with decent success.
Thankyou,
-Dianne
--
Dianne Patterson, Ph.D.
[log in to unmask]
University of Arizona
SLHS 328
621-5105
Dianne Patterson, Ph.D.