Dear Chrsitian, Lynne, Ged and the rest of the SPM community,
Thanks for your answer.
My sample aged from 12 to 17 years. There were two gropus, a control group and
a nervous anorexia patient group. The pre-post scan delay was about one year.
The control group was taken to evaluate the growth effects in VBM, in order to
remove those effects from the patient group analysis. I have still two
questions concerning how things have to be done to get the best results:
- Is modulating normalized images necessary to distinguish volume variations
due to growth from those due to the treatment applied to the patients? I think,
in this case, jacobians should be applied although this being a longitudinal
study.
- Is an ANOVA (two groups, two conditions, proportional scaling?) the best way
to distinguish which differences are related to growth from those related to
treatment?
Thanks again for your time and for your advise,
Yours
Carles
Missatge de Christian Gaser <[log in to unmask]>:
> Dear Carles,
>
> your proposed steps are ok, but I would suggest the following changes:
> - segment all scans in the first/second step without using using customized
> priors or templates.
> The impact of the priors/templates in SPM5 is minimized and you don't need to
> customize your
> priors/templates (except for e.g. children data, where the distribution of
> GM/WM/CSF deviates
> from adult subjects)
> - coregister the segmented images
> - modulation makes no sense for longitudinal data because you analyze
> relative differences
> between the scans of each subject and there is no need to correct for the
> volume changes due to
> warping, because the warping parameters are the same for all scans of one
> subject
>
> You might also try the longitudinal segmentation in the VBM2 toolbox which
> merges all these
> steps using the good old SPM2 segmentation:
> http://dbm.neuro.uni-jena.de/vbm
> The advantage of this approach is that you use the same templates and priors
> for all scans of one
> subject. In SPM5 the templates/priors might be warped differently to the
> scans, which can cause
> artificial differences in segmentation if there are large differences between
> the scans of one
> subject (e.g. atrophy)
>
> Best regards,
>
> Christian
>
>
> ____________________________________________________________________________
>
> Christian Gaser, Ph.D.
> Assistant Professor of Computational Neuroscience
> Department of Psychiatry
> Friedrich-Schiller-University of Jena
> Philosophenweg 3, D-07743 Jena, Germany
> Tel: ++49-3641-935805 Fax: ++49-3641-935280
> e-mail: [log in to unmask]
> http://dbm.neuro.uni-jena.de
>
> On Tue, 26 Sep 2006 11:53:31 +0200, CARLES M. FALCON FALCON <[log in to unmask]>
> wrote:
>
> >Dear Spmers,
> >
> >I have some doubts abouts how to do a longitudinal VBM with SPM5.
> >
> >What I was planning to do is
> >- to segment the first scans using the default priors to create my own GM,
> WM
> >and CSF priors,
> >- to segment the first scans again using my own GM,WM and CSF priors and
> >writing modulated and unmodulated normalized segmented images,
> >- to coregister the second scan to the first one,
> >- to segment the second scan and write segmented images in the native
> space,
> >- to apply the normalization parameters of the first scans, obtained in
> >segmentation step, to the second segmented scans
> >- to smooth the images.
> >In this case, to create my own template doesn't seem necessary, just the
> >priors.
> >
> >The first question is if this procedure is suitable and the most adequate.
> The
> >second question is how to change normalize writing options (voxel size)
> when
> >creating normalized segmented files from segment function.
> >
> >Any advice on these questions would be much appreciate.
> >
> >Thanks
> >
> >Carles Falcon
> >IDIBAPS-Hospital Clinic Barcelona
> >University of Barcelona. SPAIN
> >
> >
> >
> >-------------------------------------------------
> >This mail sent through IMP: http://horde.org/imp/
> >===========================================================
> ==============
>
-------------------------------------------------
This mail sent through IMP: http://horde.org/imp/
|