Hi Martin,
FNIRT is designed (for now?) to align images that have the same
contrast. FA and T1 do share some similarities in their contrasts, but
tend to behave differently when you approach the cortex. So it
probably depends what your ROIs are.
My experience with FNIRTing FA to T1 (using only default options) is
that it worked quite well for aligning obvious white matter tracts
(corpus callosum), and seemed to do a reasonable job at aligning
cortex when FA was not too reduced in the vicinity of the cortex. It
is difficult to tell how good the alignment is when you cannot see
detailed structure of the cortex on the FA image... I have tried with
2mm isotropic diffusion data, as well as 1.5mm data.
Note that if you want to do tractography using FDT, and you want to
use FNIRT, then you will need to do the tracking in native diffusion
space, as FDT does not yet allow you to input a FNIRT warpfield.
Cheers,
Saad.
On 6 Nov 2008, at 10:09, Martin Ystad wrote:
> Hello all,
> I have masks that have been generated from T1-volumes that I'd like to
> register to the DTI-domain to measure FA and ADC in the regions of
> interest
> defined by the masks (I'd also like to track fibers that pass through
> ROI's). There have been some threads on the forum earlier on this
> subject.
> From what I could gather, the best way to do this is to use FLIRT to
> register the FA image to the T1 image, and not the other way around,
> due to
> the higher resolution of the T1 image. The affine registration does
> not give
> as good results as I had hoped for, so I'd like to try FNIRT as
> well. I read
> that FNIRT was not really intended for this, but it still worked for
> some
> images. What I'd like to know: is there any consensus on how to use
> FNIRT
> for this purpose, or do I just have to try out different settings and
> visually evaluate the results? Do I specify some configuration file,
> or go
> with as few options as possible? Does anyone have experience with
> this?
>
> P.S. I'm completely new to FSL, so please let me know if I have
> misunderstood something crucial.
>
> Thanks,
>
> Martin Ystad
> University of Bergen,
> Norway.
>
Saad Jbabdi
Oxford University FMRIB Centre
JR Hospital, Headington, OX3 9DU, UK
+44 (0) 1865 222545 (fax 717)
www.fmrib.ox.ac.uk/~saad
|