Another alternative would be to combine the tissue probability maps supplied with SPM with averages you have created yourself. These averages could be smoothed slightly as well. After trying different combinations, it should then be possible to figure out what works best.
I should really explore the possibility of doing model selection within the segmentation code. This would give log-likelihoods for the different tissue probability map models, indicating the model likelihood for segmenting particular data sets. Then the model with the highest log-likelihood could be chosen.
Best regards,
-John
-----Original Message-----
From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] On Behalf Of chetelat
Sent: Friday, May 18, 2007 4:38 PM
To: [log in to unmask]
Subject: Re: [SPM] Template VBM
Dear Valentino,
According to what I've understood from previous posts:
- with SPM2: yes (use you own template)
- with SPM5: NO (use the MNI default SPM5 template) - (unless you have large
enough samples = more than one or two hundreds)?
Gaël.
-----Message d'origine-----
De : SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] De la
part de Valentino A. Pironti
Envoyé : vendredi 18 mai 2007 15:18
À : [log in to unmask]
Objet : [SPM] Template VBM
Dear all,
I am going to perform a VBM analysis with MCI subjects.
Do you suggest to create my own template using both MCI and control group?
Thanks,
Valentino
|