The -noresample flag does not stop using the reference volume
to determine the output size and resolution. It just stops internal
calculations doing resampling which, given the -applyxfm flag, is
Using your new reference image is the right thing to do.
However, the qform has nothing to do with how flirt will resample
the intensities. So if the image *looks* wrong after this then there
is a bigger problem. Getting the right qform for your output will
then be a second problem. I'm not sure if "data in the wrong
place" means that just the coordinates are wrong (incorrect qform)
or that the image looks wrong (wrong intensities, and hence a bigger
What I think you should do is the following:
- run fslorient with -deleteorient on your new reference volume,
to remove its qform info (which would otherwise taken precedence)
- run: flirt -in inplane -ref newref -applyxfm -init yshift.mat -out
Hopefully this will make an image which looks correct (the
intensities are where they should be in the image) and has the
transformed qform from the input image (as the newref will have
had no info to override this).
Let us know if you have any problems with this.
All the best,
On 4 Aug 2009, at 00:22, Ed Vessel wrote:
> Ping ... Got no response on this, just doublechecking to see if there
> are any recommendations or if I'm missing something obvious.
> On Thu, Jul 30, 2009 at 10:17 AM, Ed Vessel<[log in to unmask]> wrote:
>> Hi there -
>> I have some EPI and 2D T1 ("inplane") data that were collected with
>> the same slice prescription but different fields of view in the image
>> plane, and I'm trying to pad my inplane to have the same FOV (without
>> changing the NIFTI qform location info).
>> The EPI scan is 3x3x3mm, 64x80 voxels (192x240 FOV).
>> The 2D T1 inplane is 1.5x1.5x3mm, 128x128 voxels (192x192 FOV).
>> I've tried using some flirt commands to do this but can't get it to
>> work right.
>> flirt -in inplane -ref EPI -noresample -applyxfm -init yshift.mat -
>> ... does the right thing but the "noresample" tag seems to not have
>> effect and I get 3x3x3 voxels (not 1.5x1.5x3).
>> (note, my yshift.mat file is there just to re-center the data in the
>> middle of the slice, adding the padding on both sides rather than
>> one side. It contains a 24mm shift in y).
>> flirt -in dino_inplane_090729+03+t12Dmprage26sl -ref
>> ../04+Online/dino_inplane_090729+04+Online -applyisoxfm 1.5 -init
>> yshift.mat -out inplane_paddedFOV2
>> ... this resamples in all directions, giving me 1.5x1.5x1.5mm voxels,
>> which is no good because now I can't use it as an "inplane" ... it
>> doesn't have the same slices as the EPI.
>> I've also tried creating a _new_ reference file with the right image
>> dimensions (1.5x1.5x3, 128x160 voxels), but have been unsuccessful at
>> getting the qform of this new file set correctly, so that it places
>> the data in the wrong place in the file when I then use this as the
>> Any suggestions on how to do this?
> Ed Vessel
> Center for Brain Imaging
> New York University
> [log in to unmask]
> 4 Washington Place, Rm. 156
> New York, NY 10003 http://www.cns.nyu.edu/~vessel
> (212) 998-8217