GE scanners: single file fieldmap/magnitude

[amirhusseinab]

Hi,

I'm using Heudiconv 1.3.2 (singularity) on a dataset. I noticed subjects acquired with GE scanners have a single row in their dicominfo.tsv with series_description: "B0rf map", dims: 64x64x160. My first thought was that maybe this is "Sbref" (putting it in heuristics.py as b0rf = create_key("sub-{subject}/{session}/func/sub-{subject}_{session}_acq-rest_sbref"). After conversion, it generated two files (sbref1 and 2): one with BidsGuess: "fieldmap" with units in Hz with two Echo times and the other BidsGuess: "magnitude".

I would like to know what change should I make/add in the heuristics to have HeudiConv automatically generate these images in fmap for me.

Thanks!
Amir

[levchenkoegor]

Hey Amir,

Not sure, if it's relevant but I had a similar issue with dual-echo fieldmaps (acquired on Siemens scanner though). Unfortunately, I couldn't find the solution but there is a bit of discussion here:

#789

[amirhusseinab]

Hey Egor,

Thanks for the mention, I've seen the issue you've raised but I was hoping some fix may have been added since. Have you tried anything other than the one you told there?

[yarikoptic]

b0rf = create_key("sub-{subject}/{session}/func/sub-{subject}_{session}_acq-rest_sbref"). After conversion, it generated two > files (sbref1 and 2): one with BidsGuess: "fieldmap" with units in Hz with two Echo times and the other BidsGuess: > "magnitude".

I would like to know what change should I make/add in the heuristics to have HeudiConv automatically generate these images in fmap for me.

I must be missing the point, but shouldn't you just generate the key which has /fmap/ in it instead of /func/ since you do get two files produced, correct?

FWIW - it is dcm2niix which from a single series is extracting multiple echo times, and heudiconv just detects them and in --bids mode renames them to have _echo- entities. here: https://github.com/nipy/heudiconv/blob/HEAD/heudiconv/convert.py#L1004

[amirhusseinab]

I must be missing the point, but shouldn't you just generate the key which has /fmap/ in it instead of /func/ since you do get two files produced, correct?

So, I changed that part of the heuristics into fmap_b0rf = create_key("sub-{subject}/{session}/fmap/sub-{subject}_{session}_acq-rest_fieldmap"), and later

if ('B0rf Map' in s.series_description):
            info[fmap_b0rf].append({"item": s.series_id})

Here is the tree structure of the fmap folder:

fmap/
├── sub-01_ses-001_acq-rest_fieldmap1.json
├── sub-01_ses-001_acq-rest_fieldmap1.nii.gz
├── sub-01_ses-001_acq-rest_fieldmap2.json
├── sub-01_ses-001_acq-rest_fieldmap2.nii.gz
├── sub-01_ses-001_dir-LR_acq-rest_epi.json
└── sub-01_ses-001_dir-LR_acq-rest_epi.nii.gz

There are two "fieldmap" files, (1) is a magnitude image, (2) is a dual-echo fieldmap. The point is, in dicominfo.tsv there is only one line with "B0rf Map" for these two files. Do you have any suggestions that I should make?