Hi Bryn,
actually, this is not a difference of convention between SPM and FSL, but a mistake in the manual that we should have corrected ages ago (we thought initially that it would be easier to understand it this way, but it's simply confusing)!
If you look at the scripts, you'll see that we *multiply* the GM, normalised images by the Jacobian. So Jacobian>1 means that the structure has been contracted by the NL registration, hence was bigger than the template initially. Also note that there is a difference with SPM in that we do *not* include the affine part in the Jacobian (only the NL component), so we effectively normalise the data for brain size and TIV for instance is not required as an additional covariate (see archives for more discussion on the subject).
Hope this helps,
Gwenaelle
> Hello,
>
> We are using FSL-VBM for voxel-based morphometry analysis
> and hope to identify the volumetric changes between patients
> and normal controls. I have a few questions regarding GM
> modulation.
>
> 1. The first question is with regards to the Jacobian
> determinant, which is an output from VBM. The Jacobian
> determinent represents the relative volume between the
> subject and template. I am unsure about the
> directionality. If |Jacobian determinant| > 1, does
> it mean that the subject has a larger volume than the
> template? If |Jacobian derminant| is less than 1, does it
> mean that the subject has a smaller volume than the
> template?
>
> 2. In FSL, "the registered partial volume images were
> then modulated (to correct for local expansion or
> contraction) by dividing by the Jacobian of the warp
> field."
> However in the SPM VMB8 manual, it says the modulated GM is
> calculated as: "Affine+non-linear produces tissue class
> images in alignment with the template, but multiplies
> ("modulates") the voxel values by the Jacobian determinant
> derived from the spatial normalization.
>
> Why is modulation done differently in SPM and FSL (multiply
> vs divide)?
>
> I would really appreciate anyone's help in clarifying these
> matters.
>
> Thank you!
> Bryn
--------------------------------------------------------------------
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
--------------------------------------------------------------------
|