That's a bit more tricky as I haven't yet got around to incorporationg
Jacobian scaling into the Deformations toolbox. If you extract Jacobian
maps, then these will have different dimensions to the normalised images, but
ImCalc should (approximately) take care of this if you do a scalar
multiplication of spatially normalised images and Jacobian determinant maps
(entering .* to multiply - rather than just *).
Best regards,
-John
On Friday 12 September 2008 22:52, zewang wrote:
> But how about modulation? Should I extract the Jacobian maps and then apply
> to the normalized images? Thanks.
>
> ---------------------------------------------
> Ze Wang, Ph. D
> Center for functional neuroimaging,
> Dept of Neurology,
> Medicine school,
> University of Pennsylvania
> 3400 Spruce Stree,
> 3W Gates building,
> Philadelphia, PA 19104
> Tel: 215-662-7341
>
>
> -----Original Message-----
> From: John Ashburner [mailto:[log in to unmask]]
> Sent: Friday, September 12, 2008 12:28 PM
> To: zewang; [log in to unmask]
> Subject: Re: [SPM] how to use DARTEL for normalizing fMRI con maps?
>
> If you want to write warped images with different voxel sizes etc, then you
> can do this via the Deformations toolbox. The steps would involve
> composing
>
> the deformation field from DARTEL with an identity transform, where you can
> specify the dimensions, voxel sizes etc of the output from some other
> image.
>
> You should be able to find more hints about this in Section 1.4 of the
> DARTEL
> guide.
>
> http://www.fil.ion.ucl.ac.uk/~john/misc/dartel_guide.pdf
>
> Note that it would be slightly simpler and involve composing one Backward
> transform with an identity transform.
>
> Best regards,
> -John
>
> On Friday 12 September 2008 17:16, zewang wrote:
> > Hi John,
> > How to change the boundary and voxel size after getting the flowmap
>
> and
>
> > warping map? Do I need to resample the u_rc1 map before warping the
> > functional contrast images into the local template space?
> >
> > Thanks.
> >
> > ---------------------------------------------
> > Ze Wang, Ph. D
> > Center for functional neuroimaging,
> > Dept of Neurology,
> > Medicine school,
> > University of Pennsylvania
> > 3400 Spruce Stree,
> > 3W Gates building,
> > Philadelphia, PA 19104
> > Tel: 215-662-7341
> >
> >
> > -----Original Message-----
> > From: John Ashburner [mailto:[log in to unmask]]
> > Sent: Thursday, August 28, 2008 10:53 AM
> > To: zewang; [log in to unmask]
> > Subject: Re: [SPM] how to use DARTEL for normalizing fMRI con maps?
> >
> > Poor starting estimates could be a problem here. Try the usual stuff
>
> about
>
> > setting the origin and getting the orientation approximately correct
>
> before
>
> > you begin.
> >
> > Note that the initial affine registration does not include any bias
> > correction
> > in its model, whereas the actual "Unified Segmentation" algorithm does.
> > Therefore, if you have really good starting estimates, then you could
> > disable
> > the initial affine registration.
> >
> > Best regards,
> > -John
> >
> > On Wednesday 27 August 2008 15:01, zewang wrote:
> > > Thanks John. I figured it out by trying the batch scripts you provided
>
> in
>
> > > an email.
> > > Another question is: have you tried to segment T1 images acquired with
> > > array coil? SPM5 failed for many cases due to the inhomogeneity. I used
> >
> > FSL
> >
> > > to correct the inhomogeneity, but SPM5 still failed for couple of
> >
> > subjects.
> >
> > > Without the additionally acquired sensitivity map, what would you
>
> suggest
>
> > > for the inhomogeneity correction? (I know this question may be too
>
> harsh,
>
> > > but if anyone else has a good experience, please share)
> > >
> > > ---------------------------------------------
> > > Ze Wang, Ph. D
> > > Center for functional neuroimaging,
> > > Dept of Neurology,
> > > Medicine school,
> > > University of Pennsylvania
> > > 3400 Spruce Stree,
> > > 3W Gates building,
> > > Philadelphia, PA 19104
> > > Tel: 215-662-7341
> > >
> > >
> > > -----Original Message-----
> > > From: John Ashburner [mailto:[log in to unmask]]
> > > Sent: Wednesday, August 27, 2008 6:10 AM
> > > To: zewang; [log in to unmask]
> > > Subject: Re: [SPM] how to use DARTEL for normalizing fMRI con maps?
> > >
> > > > To use DARTEL for fMRI group analysis, where should I import the
> > > > individual con files (coregistered to the T1 images)?
> > >
> > > Providing your images are registered, then there should be no need to
> > > import
> > >
> > > your con files. Note that DARTEL hasn't really been evaluated for fMRI
> > > studies. There are likely to be a number of issues relating to how
> > > best
> >
> > to
> >
> > > deal with regional expansion and contraction, which are not properly
> > > dealt with by simple warping of the images. I've written about this a
> > > number of times, so I won't go into details here.
> > >
> > > > My understanding of
> > > > DARTEL based normalization is that there are two separate
> > > > registration processes, one for grey matter and the other for white
> > > > matter. How to combine these two processes to get a final template?
> > >
> > > That is not the case at all. Grey and white matter (and any other
> > > classes that can be well segmented) are registered simultaneously in
> > > order to obtain
> > >
> > > a consistent estimate of the deformations.
> > >
> > > > Another question is: is there a way to use this local template to
> > > > warp the new data sets that are not involved in template generation?
> > >
> > > Yes. This is what the "Run DARTEL (existing Templates)" option does.
> > >
> > > Best regards,
> > > -John
|