Hi Steven,
VBM2 seems to be still a better choice for longitudinal data. I have
implemented some additionaly features in VBM2 to preprocess longitudinal
data (two-step intra-subject registration, correction of bias between
subsequent scans) and I share your observation that the VBM2 results
outperform VBM5 for longitudinal data.
I intend to integrate some of these features in VBM8 in the future. However,
the first version will not include these features. In the meantime I would
prefer VBM2 for longitudinal data.
Best regards,
Christian
____________________________________________________________________________
Christian Gaser, Ph.D.
Assistant Professor of Computational Neuroscience
Department of Psychiatry
Friedrich-Schiller-University of Jena
Jahnstrasse 3, D-07743 Jena, Germany
Tel: ++49-3641-934752 Fax: ++49-3641-934755
e-mail: [log in to unmask]
http://dbm.neuro.uni-jena.de
On Tue, 10 Feb 2009 13:33:27 -0800, Steven Berman <[log in to unmask]> wrote:
><html>
><body>
>Dear SPM list,<br>
> We have a lovely expected result using VBM2 on a small
>longitudinal dataset (n= 3 with a very rare genetic mutation, scanned 9
>times each in 3 x 3 conditions). When we tried to reproduce in SPM5, the
>results did not make sense. We are now deciding if we should try again or
>write the paper using the VBM2 results that do. We think our results are
>real based on compatibility with previous findings, several listposts
>that cautioned about possible difficulties with SPM5 for longitudinal VBM
>(due to differences in warps within scans of an individual) and the fact
>that <u>all </u>longitudinal VBM significant published findings which a
>quick lit search turned up used VBM2 (except one paper that found no SPM5
>VBM 1-year change in 211 MS patients, but did report changes in smaller
>subsets selected for severity). This could mean VBM2 is
>oversensitive or VBM5 undersensitive for longitudinal VBM, or just
>reflect inertia. I was wondering if anyone wanted to weigh in or
>let us know about further progress that would inform the current best way
>to analyze our small longitudinal dataset.<br>
>Thanks,<br><br>
><x-sigsep><p></x-sigsep>
> Steven Berman,
>Ph.D.<br>
><x-tab> </x-tab>UCLA
>Division of Digestive Diseases, Psychiatry and Brain Research
>Institute<br>
> Center for
>Neurobiology of Stress (CNS), and Center for Addictive Behaviors
>(CAB)<br>
> Tel. (310)
>825-0616 Fax: (310) 825-0812 <br><br>
> <font color="#0000FF">
><a href="http://ibs.med.ucla.edu/" eudora="autourl">
>http://ibs.med.ucla.edu<br>
></a>
><a href="http://uclamindbody.org/" eudora="autourl">
>http://uclamindbody.org<br>
></a>
><a href="http://www.bri.ucla.edu/" eudora="autourl">
>http://www.bri.ucla.edu</a></font> <br>
> <font color="#0000FF">
><a href="http://www.semel.ucla.edu/cab/" eudora="autourl">
>http://www.semel.ucla.edu/cab/</a></font> </body>
></html>
|