Hi,
I actually think that fslswapdim is a worse way to deal with
orientation,
since you are changing the order of the acquired data, and then often
the header information as well. So after doing all of this you can
completely
lose track of what the original data was like. Hence I think that
just modifying
the header, although still somewhat dangerous for getting left/right
order
mixed up, is better than changing the data.
The best thing is to sort out your reconstruction so that this
information is
kept in the nifti images (DICOM has this info) and to make sure that all
parts of your pipeline preserve this info.
All the best,
Mark
On 6 Nov 2008, at 16:36, James E. Dobson wrote:
> Hi, Mark,
>
> Yes, indeed this was the problem. Setting the qform and sform codes
> prior to forceneurological appears to have done the trick. In the
> near future I hope to use fslswap in order to avoid this problem.
>
> Many thanks,
>
> JED
>
> --- You wrote:
> I suspect that this is your problem, although it is increasingly
> uncommon (and bad) if files are written with no legal orientation
> info, so I would check your original reconstructions and processing
> pipeline to see if this information is just never created or is lost
> somewhere along the line.
> --- end of quote ---
>
|