Print

Print


Hi Matt,
            Thanks for the response. Just to confirm, for a resampling of  V1 (say) to structural, with no registration:
flirt -in <the V1 image set> -out <resamples V1 image set> -ref <structural> -applyxfm -init <identity matrix.mat> ?
thanks,
sid.


On Wed, Dec 31, 2008 at 1:08 PM, Matt Glasser <[log in to unmask]> wrote:

Your input image will be resampled to the same dimensions as the reference image.  If you do not want any spatial transformation of the image, you can use an identity matrix as the <mytransform> in the below example.


Peace,


Matt.

 


From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf Of Siddharth Srivastava
Sent: Wednesday, December 31, 2008 2:55 PM
To: [log in to unmask]
Subject: Re: [FSL] dtifit V{1,2,3} data format

 

Hi saad,
            Thanks again. I am aware of the issue while transforming vector images. However, for my specific problem,
my DTI data and metrics are already in register with the structurals, it is just that they are of different dimensions
and voxel size. I needed to equalize for these two parameters only, and hence i wanted to reslice them to the
structurals. Can i do it in fsl, for the 4D data as input, and 3D data (T1) as target?
regards,
sid.

On Wed, Dec 31, 2008 at 10:28 AM, Saad Jbabdi <[log in to unmask]> wrote:

Hi Sid

 

If you want to transform vector images (such as V1,2,3) onto structural space, you can use the FSL command-line tool called "vecreg".

 

You need to be careful when transforming vector images, as any rotation to the image must be applied to the vectors as well (See the FDT documentation).

Note that vecreg doesn't calculate a transform for you, but uses a transform given as an output of FLIRT or FNIRT.

 

For example, you may do the following:

> flirt -in <b0image> -ref <structural> -omat <mytransform>

> vecreg -i <V1> -o <V1_to_struct> -r <structural> -t <mytransform>

 

Cheers,

Saad.

 

On 31 Dec 2008, at 18:06, Siddharth Srivastava wrote:



Hi saad,
            Thanks for the reply. since this mail was sent, i have learnt about
fslsplit and merge. I wanted to reslice/resample the metrics (V1 et al.) to my
structurals, which i do in spm. Hence i needed to exctract each component,
change them to .img, reslice, and finally merge them into the nii.gz format
(to be used by ITK SNAP). I am sure there is an easier way in FSL to do the
same thing without going through the conversion. Is there? an equivalent of
reslice/resample command in FSL?
best regards, and a happy new year!
sid.

On Wed, Dec 31, 2008 at 8:08 AM, Saad Jbabdi <[log in to unmask]> wrote:

Hi Sid,
Sorry for such a late reply!
Yes the V<i> files are in 4D format.
In order to extract the x,y,z components, you can use fslsplit.
For example:

> fslsplit outputs_V1 components

In which case the x,y,z components of V1 will be respectively:
components0000.nii.gz
components0001.nii.gz
components0002.nii.gz

Cheers,
Saad.



On 19 Dec 2008, at 00:25, Siddharth Srivastava wrote:

Hi list,
        I am sure this has been discussed before, as it is a very
basic newbie question. I am trying to find out where the 3
components of the principal directions are stored? The dtifit
command outputs _V{1,2,3}.nii.gz . is each nii.gz file in a 4d
format, having V1x, V1y and V1z components? How do i go about
extracting these components into seperate img files?
        Thanks,
        sid.

 

Saad Jbabdi
Oxford University FMRIB Centre

JR Hospital, Headington, OX3 9DU, UK
+44 (0) 1865 222545  (fax 717)
www.fmrib.ox.ac.uk/~saad

 

 

Saad Jbabdi

Oxford University FMRIB Centre

 

JR Hospital, Headington, OX3 9DU, UK

+44 (0) 1865 222545  (fax 717)