Hi Tim,

Tried your ideas, got through the first couple but when I got to

4) edit tmp.txt replacing the first number with its own negative  
(this should be easy to script if you have to do it a lot of times)

I found there was one thing in tmp.txt, the word NEUROLOGICAL - nothing
else,
No numbers. Humm , I think the qform is not set. This nii was created
with mri_convert
From freesurfer. I asked Doug Greve about it, and the rel;ated fact that
the structural data from the same subject
Came out in RADIOLOGICAL convention -- and upside down and sideways when
displayed in fslview. The data is Siemens
DICOM from an Avanto Scanner version VB 11D. Doug said the data will
come out Neurological or radiological, however
It is set at the scanner. I do not, and I asked the MR physicist, know
where this would be getting set Neurological
For the diffusion data... I am thinking this must be a common thing, to
set datafrom a modern Siemens scanner, diffusion
And structural into the same convention and orientation so they can be
registered and use the structural to guide the seed placement ?

Is there possibly better software I could use to import the Siemens data
into NIfTI. I looked a bit
And the software I found mricro etc. all seemed to have spaggetii
dependencies and such and I was
Never able to quite get it to work. I have the idea with software that
if something like load data does not work without a lot of effort
Then one should mot likely not use the software.


-----Original Message-----
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On
Behalf Of Tim Behrens
Sent: Wednesday, December 27, 2006 3:34 AM
To: [log in to unmask]
Subject: Re: [FSL] NIFTI conversion


Hi - again, I'm not sure I'll get this completely right, but I'll  
have a go.

My advice would be to ensure everything ends up in radiological  
conventions. I think this will be very important when you do  
tracking, as otherwise there will be a great deal of confusion  
between the gradient orientations and the image convention. (fslview  
always displays the eigenvectors in radiological coordinates,  
ignoring the q-form, but it displays the images in neurological if  
the q-form is set, so it can get very confusing - we will fix this  
for the next release.)

To put the diffusion data into radiological convention, you should be  
able to do the following:

1) get the qform information out of the diffusion data avworient
-getorient data > tmp.txt

2) delete all orientation information
avworient -deleteorient data

3) swap the image axes so the data are stored in radiological  
convention:
avwswapdim data -x y z data

4) edit tmp.txt replacing the first number with its own negative  
(this should be easy to script if you have to do it a lot of times)

5) reload the data with the newly relevant q-form matrix: avworient
-setqform `cat tmp.txt` data

6) check that this has also set the q-form code ( I am not sure if it  
does or not). If it doens't, set it with avworient.

7) enjoy your new radiologically stored data.

T

On 22 Dec 2006, at 20:27, Gregory Kirk wrote:

> We are experiencing a few problems with NIFTI images created from
> Siemens
> DICOM data using either MRI convert (Freesurfer) or our own in house
> software written in IDL. We want to use structural (3D T1 weighted  
> data)
> data to define ROIs for probabilistic fibre tracking. The diffusion  
> images
> are acquired in an axial plane tipped to give an approx AC-PC  
> angulation
> while the structural images are in sagittal plane, typically with  
> small
> tilts to align the slices with midline of the brain. When we  
> convert these
> images to NIFTI format and import them in to FSL we consistently get a
> mismatch between the orientation of the diffusion data and the  
> structurals
> with the diffusion data being in neurological convention while the
> structurals appear in radiological convention. The options for  
> converting
> radiological to neurological orintation don't seem to have any  
> effect (but
> changing qform does switch the slice ordering as one would expect).  
> Thus
> we were wondering exactly how FSL determines which convention to  
> display
> NIFTI images in? Obviously what we require is to get both data sets  
> to be
> displayed in the same convention so that we can register the data sets
> prior to doing the fibre tracking.
>
> Thanks in advance.
>
> Greg