protocol parameters variations

[bpinsard]

Hi there!

I am working on a very preliminary small tool for continuous multi-site/multi-vendor protocol compliance check, and the spine-generic dataset came to my mind as one of the best for testing, as it features many repeated sites/vendors/models/head-coils and protocol variations for the same model (eg. ZoomIt/no-Zoomit) with a clear open versioned protocol.

While the parameters are overall very homogeneous, I am surprised to observe variations of metadata including MR parameters between sites with the same scanner model, receive-coil and software version (while accounting for zoom-like option).

If I am not mistaken, intra-model/version variations can be observed in fields such as ImageType (eg. with or without "NORM" or "DIS*" options), ReceiveCoilActiveElements, ReconMatrixPE, and some very slight variations are also observed on EchoTime and RepetitionTime.

Do these variations come from varying versions of the protocol? Is there an easy way to get the exact version and the protocol variant (eg ZoomIt/no-Zoomit) used for each subject. Ideally, this could be included in the participants.tsv table. If these are "random" changes forced when importing the protocol into a specific scanner, maybe this should be documented too?

I also noted that pavia subjects are acquired on an unpublished "syngo_MR_D13C" protocol, was this adapted by loading another procotol? Maybe this should be documented.

Thanks!