Philips field map output by dcm2niix

dcm2niix

#1

I am using the Nov 25, 2018 version of dcm2niix to extract Philips Intera B0 map data, It produces two files: one with _ph suffix, the other with none of _e1, _e2, or _ph. Is the _ph phase difference image ? Looks like it. What is the other one ? Also, how do these two files map onto the BIDS fieldmap use cases ? None of the 4 cases described in the BIDS specs appear to really match.


#2

In my experience with Philips fieldmaps I’ve only ever gotten the ph imageas after running dcm2niix, albeit using the par/rec format which is now deprecated. however I do tend to get 3 or 4 outputs depending on the sequence the physicist programmed, in my case multiecho fieldmaps. One of them is likely the phase difference image, and the others are likely the real and imaginary components of the phase image. try taking the sqrt(e1^2 +e2^2) and see if you get your magnitude image as a result to confirm.


#3

Thanks for chiming in. I am getting two images, one _ph going from -250 to +250, the other looks like a standard EPI image. These are outputs from .dcm raw files, not PAR/REC. So this doesn’t seem to either scenario you described.


#4

Any pointers here ? I am happy to provide more details if needed.


#5

My fieldmap page includes samples to help you distinguish magnitude, phase, real, imaginary images. With regards to the scale of your Phase image being from -250…+250, you might want to try dcm2niix with -p y and -p n. Philips includes several different scaling factors (display, floating point and real world values). The NIfTI image only allows a single scaling factor. This does not matter for most MRI modalities where image intensity is relative, but for the phase map it might help to explore these. I do not have Philips equipment, but I am sure your Philips Research Collaboration Manager would be happy to help you decipher these.


#6

Thanks, Chris. I tried to investigate based on your tips. The image without any postfix ( XXX.nii ) is clearly a magnitude image. The one with “ph” XXX_ph.nii is a phase image. Going from -p y (default) to -p n makes a difference in scaling for the magnitude image in the intensity values, but almost none for the phase image (it is still -250 to 250, and there is only tiny (<0.1) intensity differences). For either image, the .json files only differ by UsePhilipsFloatNotDisplayScaling being 0 or 1, but the other 3 scale-related parameters are the same. For the phase image, they are:
“PhilipsRescaleSlope”: 0.1221,
“PhilipsRescaleIntercept”: -250,
“PhilipsScaleSlope”: 8.19,
“UsePhilipsFloatNotDisplayScaling”: 0,


#7

I no longer have access to Philips hardware nor Philips support. I would suggest you contact your Research Collaboration Manager for details.