Hi Gwen
 
I have slightly different but related question on VBM covariate.
 
Given that the Jacobian Modulation is compulsory due to the use of non-linear registration, what do you think about the idea using a covariate for intracranial volume (ICV). It seems that the jacobian handles the variability in peoples head size, and in fact does this at a local level, so as to eliminate the need for adding an ICV covariate. Is this reasoning correct?  Many VBM publications in the literature use the ICV covariate in VMB which does not make sense to me.
 
I suppose if one did an affine registration (and no Jacobian modulation is involved) then it does make sense to use an ICV covariate ? I understand the affine registration makes the voxelwise comparison in VBM invalid due to the residual erros in registration which is a whole different issue.
 
-Raj

From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf Of Gwenaëlle DOUAUD
Sent: Friday, March 04, 2011 11:53 AM
To: [log in to unmask]
Subject: [FSL] Re : [FSL] FSL-VBM covariate

Hi Kai,

1. 2. Grey matter volume is the right term, and you're right, it refers to smoothed *modulated* grey matter... Jacobian modulation is there to compensate for the "artificial" expansion or contraction of GM occurring during the non-linear registration. You need to modulate your GM images in order to be able to interpret correctly your results.

3. Please have a look at the archives for this
https://www.jiscmail.ac.uk/cgi-bin/webadmin?S1=fsl

4. Yes, it is quite common to have uncorrected differences between groups that do not survive the correction for multiple comparisons (tfce, cluster-threshold, fdr etc.). If this is the case, then I'm afraid that you don't have any significant result.

Cheers,
Gwenaelle
Dear ALL:
Hello. I write to ask you for help about FSL_VBM protocol. I am processing T1 images of two groups:patient and control. According to the classical method in the manual you write on the FSL mainpage, "the registered partial volume images were then modulated (to correct for local expansion or contraction) by dividing by the Jacobian of the warp field. The modulated segmentated images were then smoothed."
 
Then I am going to do statistical analysis.Here come my questions:
 
1. As the method description, the gray matter "volume" or the gray matter "density" are going to be compared between these two groups? What is the different between "gray matter volume" and "gray matter denstiy"? 
The following papers are done by the same method, the first paper call it "denstiy" but the second one call it "volume". I am confused.
 
2008-Neuron-The Brain in Chronic CRPS Pain: Abnormal Gray-White Matter Interactions in Emotional and Autonomic Regions 
2010-BP-Gray Matter Alterations in Adults with Attention-Deficit/Hyperactivity Disorder Identified by Voxel Based Morphometry
 
2. What is the purpose of Jacobian modulation? In my opinion:modulated refers to volume; unmodulated refers to denstiy, right?
 
3.If the patient group and the controls showed dofferent SDS, how can I include the SDS as a covariate using ANCOVA? What is the command in Randomise?
For instance:randomise -i GM_mod_merg_s3 -m GM_mask -o fslvbm -d design.mat -t design.con -T -n 5000 -V 
 
4.With regarding to the Multiple comparison issues, the TFCE corrected survied nothing, However, the uncorrected result displayed some brain regions. Is it common? What is the meaning of the uncorrected results and the significant between the corrected and uncorrected result?
 
Thank you very much.

Best wish


--------------------------------------------------------------------

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

--------------------------------------------------------------------