[log in to unmask] wrote:
> Just to make sure that I understand the difference between Jacobian
> determinants in VBM (SPM5) vs Jacobian Determinants in TBM where High
> Definiton Warping is used. Is this difference simply that when we use
> HDW in TBM we are applying larger number of cosines in basis function
Hi Shary,
Basically yes, they are the same thing, and the difference is, as you
suggest, about the flexibility/precision of the warping model.
However, I don't think it's true that HDW simply uses more DCT
coefficients. I believe it uses a different, "piecewise affine" model,
as described in chapter 4 of Human Brain Function (2nd Ed):
http://www.fil.ion.ucl.ac.uk/spm/doc/books/hbf2/pdfs/Ch4.pdf
I'm not sure of the relative merits of this piecewise affine approach
vs. more DCT coefficients, but I assume the latter has problems with
computation time and/or local optima and/or numerical stability etc.
that motivated John's development of the former.
Hope that helps,
Ged.
P.S. I should point out that I'm still not entirely certain of the
model for HDW, since I never got a reply to this message:
http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0609&L=SPM&P=R91998
but I think the basic model is probably as described in HBF2.
|