Thank you Gwenaelle. They were originally 3 groups then we were asked to do this analysis in 4 groups thats the reason of the confusion. Anyways, seems like I need to read more to get my head around this. Cheers Cagri 2012/1/9 Gwenaëlle DOUAUD <[log in to unmask]> > Hi Cagri, > > your question is clear, but, unless there is indeed a known linear > relationship between your 3 groups (???), I'm not sure I can make any sense > of what the reviewer is suggesting (1ev with 1 for controls [sic], 2 for > group X etc.). Most straightforward thing to do is what Michael Harms has > suggested in his last post, this should answer (meaningfully) to the > question of "continuum"... > > Cheers, > Gwenaelle > > -------------------------------------------------------------------- > Gwenaëlle Douaud, PhD > FMRIB Centre, University of Oxford > John Radcliffe Hospital, Headington OX3 9DU Oxford UK > Tel: +44 (0) 1865 222 523 Fax: +44 (0) 1865 222 717 > www.fmrib.ox.ac.uk/~douaud > -------------------------------------------------------------------- > ------------------------------ > *De :* cc yy <[log in to unmask]> > *À :* [log in to unmask] > *Envoyé le :* Lundi 9 Janvier 2012 15h15 > *Objet :* Re: [FSL] Re : [FSL] Kernel size (VBM) > > Thank you Gwenealle. > > Yeah, I will choose the safe way and make sure the reviewer feels > satisfied :) > > I know that you are knowledgable about FSL. I wonder if you have read my > previous question, which I couldnt find a final answer for. I am copying it > in case you haven't. I would very much appreciate any contribution. (our > all discussion is under the topic 'Multiple regression model with a group > variable (VBM)<https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=fsl;29cd0501.1201> > ') > > Best > > Cagri > > Dear FSL experts, > > I am reposting this question with the hope of making it more clear this time. > > We have 130 subjects in 4 groups (3 patient groups X,Y,Z and a control group) and we compared the GM volumes between these groups in an ANOVA design. > > We were asked by a reviewer to do a 'multiple regression with a group variable (e.g. X=4, Y=3, Z=2, Controls=1)' and, 'an independent estimation of group variables' to examine the continuum of GM abnormalities in this sample. > > I am copying the exact words of the reviewer to give a better idea about what is asked from us (I just changed the group names): > > " the examination of a potential continuum of abnormalities between these groups is likely the more adequate approach. Given the limited statistical power of this study this is probably best examined in the context of a multiple regression model with a group variable (e.g., Group X = 4, Group Y = 3, Group Z = 2, healthy volunteers = 1) and an independent estimation of (potentially unequal) group variances. Given the strong a priori evidence for a continuous increase in gray matter deficits in prefrontal and temporal cortices over groups, one-sided testing of such a model appears legitimate." > > How should the design matrix be to test this? > > Thank you very much > > Cagri Yuksel > > > > > 2012/1/9 Gwenaëlle DOUAUD <[log in to unmask]> > > Hi Cagri, > > alternatively, you can simply do: > > fslmaths GM_mod_merg -s 5.1 GM_mod_merg_fwhm12 > > and then run randomise on GM_mod_merg_fwhm12 ... > > A larger kernel size means that you'd pick up differences that are larger > between your two groups. Unless you had possibly two small peaks that > merged into a big one by changing the kernel size, if this was not "there" > at 7mm, there's a good chance it won't "appear" at 12mm, but it's always > better to please the reviewer and just do it :-) > > Cheers, > Gwenaelle > > -------------------------------------------------------------------- > Gwenaëlle Douaud, PhD > FMRIB Centre, University of Oxford > John Radcliffe Hospital, Headington OX3 9DU Oxford UK > Tel: +44 (0) 1865 222 523 Fax: +44 (0) 1865 222 717 > www.fmrib.ox.ac.uk/~douaud <http://www.fmrib.ox.ac.uk/%7Edouaud> > -------------------------------------------------------------------- > ------------------------------ > *De :* Cagri Yuksel <[log in to unmask]> > *À :* [log in to unmask] > *Envoyé le :* Dimanche 8 Janvier 2012 16h34 > *Objet :* Re: [FSL] Kernel size (VBM) > > Thank you very much Anderson, just started FSLVBM_3_proc with sigma 5, > hopefully it'll work. > > Well, in our study we compared 3 groups and we couldnt find difference > between two. The reviewer suggests that the small Kernels are better for > subcortical structures and larger kernels for cortex (which is actually in > contrast to what I had read in a recent review paper). So he recommends > using a larger Kernel to reveal the possible cortical GM differences > between those 2 groups, which we couldnt find with 7 mm. > > Best regards > > Cagri > > On Sat, 7 Jan 2012 21:18:03 +0000, Anderson Winkler < > [log in to unmask]> wrote: > > >Hi Cagri, > > > >This can be changed. Perhaps the easiest is if you open the file > >${FSLDIR}/bin/fslvbm_3_proc and change the line 103 (for FSL 4.1.9), > >adding or removing sigmas you want: > > > >* for j in 2 3 4 ; do* > > > >If you want, for instance, sigmas = 5 and 8 mm, just add: > > > >* for j in 2 3 4 5 8 ; do* > > > >Then run again the pipeline from this script onwards. > > > >I wonder, though, what would be the rationale for the reviewer's > >request, as we often wish to have less smoothing... > > > >Hope it helps! > > > >All the best, > > > >Anderson > > > > > > > >On 07/01/12 16:01, Cagri Yuksel wrote: > >> Dear FSLers, > >> > >> We were asked by a reviewer to repeat our VBM study with a FWHM > kernel size of 12 mm (we reported results in sigma 3mmx2.3=7 mm). However > as far as I know from the FSL-VBM website the highest kernel size FSL > allows for VBM analysis is 9 mm (sigma 4x2.3). > >> Is there a way of doing the VBM analysis with FWHM 12 mm kernel ? > >> > >> Thank you, > >> > >> Cagri Yuksel > >> > >> > > > > > > > > > >