Hi Rick,

Thanks for replying.

rick reynolds wrote:

> Hi Colm,
>
> I don't think there are any problems here.
>
> Identical datasets should look identical in AFNI, regardless of
> which of the 48 orientations they happen to have. AFNI makes a
> sagittal image look sagittal. While 3dresample may change the
> order of the data on disk, it should look the same in AFNI.
>

Does that mean that right of the brain should be on the left of the screen and vice versa? (I let to3d figure out the orientation of the data from the IMA files itself.)

> ---
>
> FSL should not require LPI data when dealing with NIfTI
> datasets
> (it would when using ANALYZE datasets). So the 3dresample
> command,
> while not harmful, may not do anything useful, either.
>
> ---

The effect of it is that the data shows up correctly in fslview I was told on the FSL list that the data has to look like the good jpg and not the bad jpg attached.

>
> I know little of how FSL evaluates a dataset as being
> NEUROLOGICAL,
> though I can guess (right-handed, maybe). It may even have
> some
> user interface variable to flip the image in the viewer.
>

From the FSL FAQ:
Our technical definition of "neurological" convention for image storage is that the mapping between the voxel-coordinate-system and the nifti-real-world-coordinate-system (i.e. the sform or qform matrix) has a positive determinant. Note that the standard images, the avg152T1 images, are all in "radiological" convention as was the convention for most Analyze images. The avg152T1 image is now our definition of "radiological", and we now strictly enforce this convention for Analyze images (see previous question)."

I've long since forgotten what the determinant of a matrix is

> In any case, I suggest you rely on what is left in AFNI, and
> then
> visually verify that it is on the same side in FSL.
>
> - rick
>

I took your advice and looked at them in afni and fslview and they look the same. I think I'll give at go with the one that makes fslorient say "radiological" and see what happens. It's just 2 days of compute time.

Bye,

Colm.