Dear Wayne,

The warning is there precisely so that it is clear that a reflection,
rather than just a rotation, is occurring.  This makes sure that
whenever the left-right nature of the image is changed, it is very
obvious.

When FSL reads Analyze files it assumes a certain left-right
convention (since this cannot be stored inside an Analyze file).
If you use Nifti instead of Analyze then you can correctly store
all of this information.  We always save a negative pixdim in the
Analyze files - but it does *not* mean anything about left-right
conventions (we will read it the same regardless of the sign of
this pixdim).

So for your problem I would be particularly careful that it has
not changed the left-right nature (look at the labeling in FSLView
on an image where you can visually tell left from right and check
that it is correct).  If it is correct then what you are doing is fine
and you just ignore the warning.  However, you might want to
do -x -y z, instead of just x -y z, so be very careful when you are
checking this, as only one of them will have the left-right around
the correct way.

With respect to your gradient directions, I imagine that you
will have to multiply all the y components by -1, but it will
depend if they are stored consistently with your image reconstruction
or not.  If you do a tensor reconstruction and look at the principle
directions in FSLView then you should be able to tell if the gradient
directions are correct or not based on whether the arrows line up
correctly or not.

I hope this helps.
All the best,
	Mark



On 21 Apr 2006, at 00:09, Wayne Su wrote:

> Hello,
>
> I have several sets of Philip 3T DT-MRI data. In the PAR file,  
> there are:
> .    Patient position                   :   Head First Supine
> .    Preparation direction              :   Anterior-Posterior
>
> r2aGUI MATLAB code from SourceForge is used to convert the raw data to
> Analyze 7.5 format. When I load this DT-MRI data into the FSLView, its
> Anterior-Posterior direction is different with Radiological  
> convention. So I
> use the 'avwswapdim infile x -y z outfile' to flip the y direction. It
> shows:
> WARNING:: Flipping Left/Right orientation (as det < 0).
> I looked the headinfo using 'avwinfo' and found the pixdim1 is  
> negative.
> However, I am sure the left/right orientation is correct. So how I  
> can get
> rid of this warning.
>
> Another concern is since I flipped the y direction, should and how  
> I change
> the diffusion gradient directions (b-vector) so I can get the correct
> diffusion tensor data?
>
> Please help me out these issues, and your reply will be appreciated.
>
> Best regards,
> Wayne
>
> -- 
>
> Wayne Su
> IT Coordinator
>
> RM211-828 W 10th Ave
> VGH Research Pavilion
> Centre for Complex Disorders,
> Vancouver, BC
> V5Z 1L8
>
> Tel: 604-875-4111 ext 61705
> Fax: 604-875-4376