The following green quoted text comes from another topic about general import error. I submit this this new topic to discuss more precisely about confusing orientation after converting DICOM to ANALYZE or NIFTI.
I find a bug in the previous (and actually the newer 3.2.0) AimsFileConvert version from DICOM to NIFTI: the output NIFTI is "NEUROLOGICAL" oriented in the corresponding field anyway the orientation is actually RADIOLOGICAL. It's quiet confusing. Using the same binary, the conversion from DICOM to ANALYZE is OK for the orientation field ("RADIOLOGICAL"). Additionaly, if the DICOM images are acquired from TOP to BOTTOM, the Z orientation is wrong after DICOM>NIFTI or DICOM>ANALYZE using AimsFileConvert (tag Sup for BOTTOM)
Denis Riviere answered :
(I guess it's better to start a new topic for a different problem, but never mind)
The Dicom reader in Aims is clearly incomplete and had a lack of maintainance for years, and I am not really surprised that it doesn't handle orientation correctly (it probably doesn't take care of it at all). Fabrice has recently made some changes and improvements in it, as well as added writing support for Dicom. But I don't know if such an orientation problem is fixed or not. Moreover it doesn't really explain why Analyze and Nifti formats behave differently... or maybe yes after all but I'm not sure right now before investigating further...
I just hope you will be able to run the 3.2 version to test this and report again if it still fails, because orientation problems are always incredibly complex to track and I don't want to get a headache if the problem has been fixed since then...
Hi Sebastian,
Well, I'm not the Dicom expert, and I'm afraid I can't do much without at least the help of Fabrice. I can ask him if he wants to look at it...
But I still wonder why it doesn't produce the same result when writing Nifti and Analyze. If I can access one of your images somewhere I can at least look at this issue...
Denis
Hi,
I had a look at your image.
Yes it is not saved to disk in physically the same orientation in Analyze and Nifti formats, but this should not be a problem: Analyze is saved (by default) in radiological orientation because SPM interprets it this way by default.
Nifti is saved here in neurological orientation, I don't actually remember if it is the standard default, but as the image orientation is really stored in the Nifti file header, it is not a problem: every Nifti-compatible soft should take it into account.
I have checked that Aims/Anatomist and Mricron read it in the same orientation. Both can convert to Nifti, and both niftis are read and viewed the same way in both software (even if the physical orientation on disk may be different, but it's an internal format storage convention, the "real world" is the same).
I have also checked in FSLview that both images are read and displayed accordingly.
However I don't see the Z axis flip on this image - I guess you have the problem on a different dataset.
So I would say: everything is OK on this image...
Denis
