Hi,
This is all correct. Although to get proper standard coordinates
you'll need to get the sform set correctly. Using ApplyXFM won't
do this by default. You are better off making an empty volume of
the required size and resolution with avwcreatehd, then resampling
the avg152T1 image using the empty image as the -ref in flirt (if you
omit -init it will use an identity transform, which is what you want).
At that point, the output image should be an upsampled version
of the avg152T1 with the right resolution, FOV and sform. You can
then use this for registration of your image.
You can also use Colin but, as Ged says, the results may be a little
different and biased by the idiosyncrasies of this individual brain, as
opposed to using the average template. The version of Colin from
mrc-cbu.cam.ac.uk might also have left-right swapped, as we've
had a previous report on the list pointing out that it was inconsistent
with the icbm_view version (in terms of coordinates and hence left/
right).
This was due to the ambiguity in left-right for Analyze images, as
they have chosen the opposite convention that FSL uses. So if you
do take these images, check carefully, as they may have fixed this
at their end (we did request it) or not.
All the best,
Mark
On 21 Feb 2007, at 11:39, Ged Ridgway wrote:
> Hi Sandy,
>
>> Does anyone have a recommendation on how to put a high res MRI
>> into MNI space with FLIRT; without having to downsample to the MNI
>> brain resolution, but to keep the high resolution of the MRI?
>
> I don't think you can do this in a single call to flirt/Flirt, but
> after you have used FLIRT and stored the resulting transformation,
> you can use the ApplyXFM GUI to transform your image with either
> specified output voxel dimensions or to match a target volume (e.g.
> Colin27, as David suggests; you might get slightly different
> results registering to avg152 and Colin27, even though they are in
> the same space. Since FLIRT has been extensively tested with avg152
> that might be the best bet...)
>
> Hope that helps,
> Ged.
|