Hi Marko,
Sorry for the confusion, I wasn't suggesting that the approach of only using a subset of subjects worked because of a eventual group difference. I ran the pre-processing using all subjects. Then looked as to whether there was still a group difference in the lesioned area. The reason that I was initially thinking about using only a subset from creating the DARTEL template is that I was worried that with unbalanced groups the template would not show any lesion and that as a result the normalization of the patients individual scans might be biased. That still might be the case but at least the lesion is still very clear in a VBM comparison. (so I'm not claiming that my original idea was valid, I haven't even tried it out yet since I'm inclined to agree that it might not be good practice to pre-process your data differently for different subjects).
With resliced I mean that during the DARTEL normalization procedure I resliced my scans to a slightly smaller voxel size (from 3mm to 2mm isotropic). I indeed remember that I read somewhere earlier that the wavy lines might have something to do with aliasing. Would smoothing the data fix that? And what would the effect be on individual voxels time-courses?
Cheers,
Richard
________________________________________
From: Marko Wilke [[log in to unmask]]
Sent: 21 March 2013 16:10
To: Richard Bethlehem; [log in to unmask]
Subject: Re: [SPM] DARTEL normalization functional lesion scans, small sample
Richard,
> Thanks for your suggestions! I will probably try to mask out the
> lesion prior to smoothing. I have just finished the 'standard'
> DARTEL procedure described earlier and it seems that the lesion is
> still apparent in a VBM group comparison thus I am quite confident
> that the 'unbalanced' group design has not diminished the lesion.
I would claim that you cannot tell from the images whether the approach
is valid. It is a statistical problem, not one of data processing, if
you only use some of your subjects to generate the template you then
normalize all subjects to.
> It appears as though there are some wavy lines in my resliced and
> normalised functional images (I'm not sure if this is fully tackled
> by subsequently smoothing the scans). Is this something that is seen
> more often? Is it related to using the DARTEL
Yes (see the archives for some explanation by John which, if I remember
correctly, has to do with pulling and pushing data). Not sure what you
mean by "resliced and normalised" - in order to be normalized, your data
will have to be resliced anyway.
Cheers,
Marko
--
____________________________________________________
PD Dr. med. Marko Wilke
Facharzt für Kinder- und Jugendmedizin
Leiter, Experimentelle Pädiatrische Neurobildgebung
Universitäts-Kinderklinik
Abt. III (Neuropädiatrie)
Marko Wilke, MD, PhD
Pediatrician
Head, Experimental Pediatric Neuroimaging
University Children's Hospital
Dept. III (Pediatric Neurology)
Hoppe-Seyler-Str. 1
D - 72076 Tübingen, Germany
Tel. +49 7071 29-83416
Fax +49 7071 29-5473
[log in to unmask]
http://www.medizin.uni-tuebingen.de/kinder/epn/
____________________________________________________
|