search

bids-standard / bids-specification

Brain Imaging Data Structure (BIDS) Specification
https://bids-specification.readthedocs.io/
Creative Commons Attribution 4.0 International
275 stars 157 forks source link

BEP-001 clarification thread #721

Closed effigies closed 3 years ago

effigies commented 3 years ago

I'm working on getting some BEP-001 example datasets in shape prior to the spec release. As I have questions I'll post them here. Some of them may require spec clarification, so I'm opening the issue here rather than in the examples repo.

1) Working on ds-mp2rage, the current MP2RAGE files have the inv entity, but not flip. This makes sense to me in that they co-vary, so either could have equally validly been chosen. That said, the template here declares inv and flip are both required for MP2RAGE. That does not seem to match anything else in the spec, though. Is this a problem with the template?

2) ds-mtsat initially had acq-T1w, acq-MTon and acq-MToff. I changed these to flip-2, flip-1_mt-on and flip-1_mt-off. Was that correct? We'll need to add metadata, but that can go in examples, as long as we get the baseline issues done here.

More questions to come as I come up with them.

cc @agahkarakuzu @Gilles86 @ChristophePhillips (please add other BEP-001/qMRI people I've missed) @bids-standard/maintainers

Converted datasets for your approval:

Gilles86 commented 3 years ago

Hey @effigies

Great to hear that you are working on this!

The central idea of MP2RAGE is that you acquire largely the same image for two different inversion times that are hence T1 and PD-weighted.

For effiency reasons, these two images usually have different flip angles, but they absolutely don't have to (whereas by definition they have to have different inversion times). I would, thus, like in my example datasets on osf, only use the inv- key/value pairs in the file names. I cannot think of any use case where it makes sense to me to use the flip- key/value pair in the file name of MP2RAGE files, but maybe one of the other cc'd people can!

-Gilles

effigies commented 3 years ago

Okay, it looks like we need to remove flip from being a mandatory entity in the MP2RAGE template. This is probably in the schema.

agahkarakuzu commented 3 years ago

That's a great catch, thanks @effigies. The flip in the MP2RAGE scheme should have been a leftover that slipped through the cracks. What you started on https://gin.g-node.org/markiewicz looks great, sorry I've been MIA due to other occupancies, should I contribute on G-Node when I can?

effigies commented 3 years ago

You can do that or comment here. If this looks good enough to move to a bids-examples PR, I'm happy to move there, which might be a little easier.

effigies commented 3 years ago

For RepetitionTimeExcitation and RepetitionTimePreparation, these are placed under Anatomy Imaging Data. These apply generally to MRI, correct? For example, this is defined in ds-mpm/sub-01/fmap/sub-01_acq-bodyMTw_RB1COR.json.

agahkarakuzu commented 3 years ago

@effigies in absence of RepetitionTimePreparation, RepetitionTime and RepetitionTimeExcitation are interchangeable, and yes, both apply generally to MRI. It is just that the current definition of RepetitionTimeExcitation is closer to the general MRI more than the current RepetitionTime is.

Is it feasible to accept both RepetitionTime and RepetitionTimeExcitation (when there is no RepetitionTimePrep) valid until there is a more radical change about this in, let say, v2.0 ?

effigies commented 3 years ago

Moving discussion to bids-standard/bids-examples#245.

I think we can push off expanding RTE/RTP to non-anatomical data in a backwards-compatible way to the next release. Nothing stops people from putting the metadata in unofficially for the moment.