search

nipy / heudiconv

Flexible DICOM conversion into structured directory layouts
https://heudiconv.readthedocs.io
Other
236 stars 126 forks source link

CMRR-MB bold magnitude/phase in separate series #518

Open bpinsard opened 3 years ago

bpinsard commented 3 years ago

Summary

We use the CMRR multiband sequence for BOLD functional scans, and recently activated the export of phase data. It results in 1 series for the SBRef which includes 1 magnitude + 1 phase image, but the accelerated BOLD volumes are exported as 2 separate series: 1 for magnitude and one for phase.

With our heuristics, that was derived from reproin one a few years back, it results in the phase series being labelled as a duplicate, while the SBRef is appropriately converted with part-* BIDS tag (it triggers the update_complex_name function as expected).

I was wondering if this is something that happened to others and if the reproin heuristics should be fixed for that? A while ago, I looked into custom grouping of dicoms to see if I could group the 2 accelerated BOLD series in one, such as to maybe trigger the udpate_complex_name function, but weirdly not many base DICOM tags are matching between the 2 series.

I was able to hack our heuristics to look if it's phase data and then modify the magnitude and phase data to include the part-* BIDS tag. It then outputs a template events file for both magnitude and phase, but this is not a big deal.

Let me know what you think. Thanks!

Platform details:

Choose one:

oesteban commented 1 year ago

I guess not having 'M' and 'P' in the image_type metadata doesn't help here (our CMRR installation does not seem to generate them).

Any news on this particular? @bpinsard @yarikoptic @mgxd

yarikoptic commented 1 year ago

oh, it has been awhile with us not looking into it, sorry @bpinsard . Does this issue still pertains and either you would have some sample files to share to troubleshoot on? might indeed be some missing DICOM information, worth reporting to either CMRR or Siemens or DICOM folks (yet to figure out who)?

oesteban commented 1 year ago

I can share some DICOMS I guess - how's the preferred way @yarikoptic ?

yarikoptic commented 1 year ago

Depending if private or public, but anything works for me. If public, could even publish datalad dataset to eg OSF;-)

oesteban commented 1 year ago

It's my brain, so it can be public. Is there anything it's better I anonymize?

On Mon, Jun 26, 2023, 16:58 Yaroslav Halchenko @.***> wrote:

Depending if private or public, but anything works for me. If public, could even publish datalad dataset to eg OSF;-)

— Reply to this email directly, view it on GitHub https://github.com/nipy/heudiconv/issues/518#issuecomment-1607664449, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAESDRR6ZJGPXRGWNFIJXNLXNGPRRANCNFSM5ARHOKAQ . You are receiving this because you commented.Message ID: @.***>

yarikoptic commented 1 year ago

it is up to you. I am not sure if there is defacing tool which operates on dicom, is there?

oesteban commented 1 year ago

Defacing is fine, and it's BOLD. I was thinking of age and stuff like that.

On Mon, Jun 26, 2023, 22:52 Yaroslav Halchenko @.***> wrote:

it is up to you. I am not sure if there is defacing tool which operates on dicom, is there?

— Reply to this email directly, view it on GitHub https://github.com/nipy/heudiconv/issues/518#issuecomment-1608238712, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAESDRXDNYJ6IVNJKENOB7DXNHZBBANCNFSM5ARHOKAQ . You are receiving this because you commented.Message ID: @.***>

bpinsard commented 1 year ago

Sorry for the late reply, I was away for the month of June. Here is a related issue I opened on the CMRR seq https://github.com/CMRR-C2P/MB/issues/305 for DWI which has been fixed in the most recent pre-release. On our side, we are still using the custom heuristics which adds the part- entity to the files.