questions > dcm2niix = how to determine blip direction from .json file
Showing 1-3 of 3 posts
Results per page:
Jun 28, 2020  10:06 AM | Kathleen Hupfeld - University of Florida
dcm2niix = how to determine blip direction from .json file

I collected field map / resting state MRI data and converted from DICOM -> NII using dcm2niix. I am having difficulty interpreting the "blip direction" for my rs-MRI data based on the .json file produced by dcm2niix. 

The rs-MRI images have "PhaseEncodingDirection": "j" in the json file. This matches my P>>A field maps (j direction), and not my A>>P field maps (j- direction). 

I need the rs-MRI blip direction for use with the SPM FieldMap toolbox: 
-Toolbox manual: The VDM is multiplied by +/-1 to indicate whether the K-space traversal for the data acquisition has a +ve or -ve blip direction. This will ensure that the unwarping is performed in the correct direction. 

However, I wasn't sure how the PhaseEncodingDirection maps onto +ve vs. -ve blip direction? (i.e., does j = +ve and j- = -ve? Or is there anything more complex that I am missing?) 

There is some conflicting information online that suggests relying on visual inspection wrt blip direction
-Oregon LCNI: Polarity of the blips depends not only on your acquisition but also on how your data was converted to NIFTI, so it could be either + or - ve ( 
-Harvard: Blip direction: -1 (but really this partly depends on how you convert your dicoms, so for your first subject, you should try it both ways and see which way is better (

Any insight would be greatly appreciated. 

Jun 28, 2020  10:06 AM | Chris Rorden
RE: dcm2niix = how to determine blip direction from .json file
This sounds like a question for the SPM mailing list, and not the dcm2nii forums. dcm2niix will save the tag PhaseEncodingDirection as specified by the BIDS standard. For an axial scan acquired on a Siemens scanner, one would by default expect an A->P acquisition to report
  PhaseEncodingDirection": "j-",
and a P->A
  PhaseEncodingDirection": "j",

Most will want to ignore this paragraph, but for completeness.... By default dcm2niix will flip the order of rows as written to disk (isFlipY = TRUE). The reason for this is that by default DICOM images are written with rows starting at the top left corner, just like you read English text. In contrast, the NIfTI format origin is the bottom left, with subsequent rows ascending the same way we draw the "y" coordinate in Cartesian graphs (y-positive = up).  This would be reversed if the rows are written to disk in reverse order (the undocumented arguments '-y n' will generate flipped rows, and the conversion will warn you Advanced feature: You are flipping the default order of rows in your image. I would expect someone would only flip this if they are trying to emulate the output of some other tool. Switching the isFlip option changes how the image data is saved to disk, and the corresponding coordinates in the NIfTI header (SForm, QForm), bvecs and BIDS (PhaseEncodingDirection). 

The SPM Fieldmap toolbox predates the BIDS standard. I am not involved in its design and have no insight. You could always run both options and see which does the best job.
Jun 28, 2020  11:06 AM | Kathleen Hupfeld - University of Florida
RE: dcm2niix = how to determine blip direction from .json file
Great, thanks so much for the quick reply! This is a helpful start to teasing out this issue. Will post on the SPM list for further clarification on the fm toolbox.