There are currently no rules regarding how to express the intent of RF fieldmaps (B1+/B1-). I don't actually know if B1 field maps are typically paired to individual structural images or not, so this may not be a problem (though we should be explicit in the text if that's the case). If it is useful to link B1 field maps to target images, then we may want to introduce something like B1PlusFieldIdentifier/B1MinusFieldIdentifier and B1PlusFieldSource/B1MinusFieldSource.
This came up in #2183.