search

moloney / dcmstack

DICOM to Nifti conversion with meta data preservation
Other
72 stars 51 forks source link

original dicom voxel order #46

Open romainVala opened 7 years ago

romainVala commented 7 years ago

Hello

I use dcmstack to convert the dicom with the .to_nifti function with voxel_order='LAS' but I get into trouble with diffusion data to get the correct diffusion direction. So for those cases I would like to write with the same orientation as dicom. So I put voxel_order='' and try to convert a sagital acquisition. The voxel order I expect (and the one give by dcm2nii) is : [ 3 -1 2 4 ] but I obtain with dcmstack : [ -3 -2 -1 4 ] do you know why and if it is possible to make the order same as the acquisition ?

many thanks

Romain

moloney commented 7 years ago

I am not sure how dcm2nii produces its result, but we do skip any kind of data re-ordering when you pass voxel_order=''. Of course the sign of the slice direction is somewhat arbitrary since the slices are not stored in a contiguous array in a DICOM. I am pretty sure dcmstack will just choose the right-handed coordinate system in this case.

If you are convinced something else is going on here, please let me know.

romainVala commented 7 years ago

Hi, You may be wright, dcm2nii may rotate the plan because there is a 90° in plane rotation between the 2 conversions. but I do not know which on is more similaire to the dicom voxel order. Any way my concern was to end up with diffusion direction coherent with the pixel order, since I was using dcm2nii as ground truth ... May be the correct diffusion direction will be output in a near futur ...