So, to be perfectly clear, the problem is *not* with the FreeSurfer tools. The problem is with fslview. While it reads the orientation information, it does not act upon it in order to reorient the images in a reasonable way. FSL has rather strict requirements on how the pixels must be arranged in order to use the tools (see www.fmrib.ox.ac.uk/fslfaq) -- this is above and beyond simple left-right orientation. Interleaved axial images from a Siemens scanner will, by default, not be arranged correctly. FSL provides tools to flip and reorient the data (this cannot be done with mri_convert). Try viewing the volumes in tkmedit (which is geometry-aware). If tkmedit displays them incorrectly, I will happily fix the problem.

doug

On Wed, 15 Nov 2006, David M Schnyer wrote:

Well, good point. When I bring it into FSLview the brain is not oriented as expected. For example the lower left window has axial scans that are upside down. Interestingly, the labels show the correct directions since A is pointing down and P is pointing up but I have run into trouble in the past running images like this through the FSL stream. Anyway, I will put these images somewhere local where you can get at them when I get in tomorrow. thanks,

david

Doug Greve wrote:

...

How do you know it's the wrong orientation? Is the data local to the our network?

On Wed, 15 Nov 2006, David M Schnyer wrote:

...

I have been trying out the dev version of mri_convert on some GE dicom images. This is functional data. If I leave the orientation alone and just convert from Dicom to nifti then I get a timeseries volume in the wrong orientation. OK, so I add the flag --out_orientation RAS. Here I get a volume in the right orientation but the timeseries is gone. Now, it isn't really gone, in that the file is the correct size but FSL no longer sees it as a timeseries. Anyone dealt with something like this?

thanks,

david