I am using dcm2niix ( latest-version ) for gre_field_map (IMA) collected on Siemens Prisma. I was hoping for two-magnitude images, a phase image as the output. However- I am getting only one "-e2_ph.nii "
Any suggestion?
If I am not wrong, we need at least one magnitude and 1 phase image for fMRIPrep to perform distortion correction ?
You can find sample Siemens Prisma fieldmaps here. Feel free to Phoenix the DICOM onto your scanner to clone it (thought remember to adjust the field of view, slices, resolution to match your other sequences). Note that the images that Siemens saves depends on your Reconstruction setting. If you want to use FSL’s Fugue you will need to select saving both magnitude and phase images:
Dear Chris,
Thank you so much for your quick response. To make sure, I double check my old data where fmap were processed using V1.0.20180622 verson of dcm2niix. I can see two-magnitude and one phase images there. However, when I ran on same data with recent version of dcm2niix ( V1.0.20211006) just produces the “e2_ph_nii.gz” .
Thank you so much.
Brain
Feel free to send a link to the DICOM data (e.g. zipped archive on DropBox or Google Drive) to my email address.
1 Like
@Brain if you plan to use Siemens Gradient Echo fieldmaps for use with BIDS or FSL’s Fugue, it is crucial that the images are acquired with the Reconstruction set to store both magnitude and phase. This must be done on the scanner console prior to acquisition.
The dcm_qa_fmap repository includes series acquired to store Magnitude/Phase (Series 2,3; CSA ucReconstructionMode 8), phase only (4, 2), and magnitude only (5, 1).
Looking at your sample dataset with a hex editor, the CSA header reports ucReconstructionMode = 2. This means that the sequence was set to store phase data only, and like series 4 in dcm_qa_fmap only a single volume is stored (I thinks this reflects a difference phase map). The difference in behavior you observe reflects the acquisition settings, not the version of dcm2niix used. If you only request a phase map, the scanner will not store a magnitude image. The data you acquired is insufficient: you must either have a magnitude image or complex data that can be used to generate a magnitude image.