[BUG] Entity requirements for MPM suffix do not appear correct

[lbanellis]

Describe your problem in detail.

page in the spec https://bids-specification.readthedocs.io/en/stable/appendices/file-collections.html we tested with @melanieganz @CPernet @Remi-Gau MPM it is valid only if entities flip and mt are present

• issue 1: in the table it does not explain what are the required entities
• issue 2: in our case we use multiple echo times and single flip, makes no sense one entity over others matter more

Describe what you expected.

No response

BIDS specification section

https://bids-specification.readthedocs.io/en/latest/...

[lbanellis]

issue one can be fix right now, then we can disccuss issue 2 how to solve it @agahkarakuzu

[Remi-Gau]

Relevant schema section:

https://github.com/bids-standard/bids-specification/blob/3bf993b1d5d40e87375f554f0214d321e9696677/src/schema/rules/files/raw/anat.yaml#L176-L198

[agahkarakuzu]

@lbanellis the entity table explains required entities. As well as the suffix:

issue 2: in our case we use multiple echo times and single flip, makes no sense one entity over others matter more

It makes sense in the context of an MPM acquisition, which is why it exists as a file collection. For example, if you are not varying your FAs, how do you get an R1 map? If you are not, is that an MPM?

[CPernet]

@agahkarakuzu what we meant is it makes no sense the echo is not mandatory and flip is, given that here the parameters that matters is echo time, ie what is mandatory or optional is weird (fixed echo = don't mention it but of course you have multiple flips, conversely fix flip = required to write flip-1 for the different echos)

[CPernet]

we have to do

sub-X_acq_PWw_echo-1_flip-1_mt-off_MPM.ext  
sub-X_acq_PWw_echo-2_flip-1_mt-off_MPM.ext  
sub-X_acq_PWw_echo-3_flip-1_mt-off_MPM.ext  

while

sub-X_acq_PWw_flip-1_mt-off_MPM.ext  
sub-X_acq_PWw_flip-2_mt-off_MPM.ext  
sub-X_acq_PWw_flip-3_mt-off_MPM.ext  

is fine

[CPernet]

issue one is to have the info in the table

[lbanellis]

Also, noticed the BIDS validator is very sensitive to the order of the entities (is only happy when echo and run is after acq)

[Remi-Gau]

Also, noticed the BIDS validator is very sensitive to the order of the entities (is only happy when echo and run is after acq)

That's expected: entities can only appear in a specific order.

The order when taking ALL entities into account is the one you can see in this document: https://bids-specification.readthedocs.io/en/latest/appendices/entities.html

More succintly put in the schema: https://github.com/bids-standard/bids-specification/blob/3bf993b1d5d40e87375f554f0214d321e9696677/src/schema/rules/entities.yaml

[lbanellis]

Thank you for your replies, this issue was written with the organisers of the BIDS BrainHack event as it took us some time to convert the MPMs to validated BIDS with the organisers help.

[Remi-Gau]

@lbanellis

Was pointed out to me that sending links to yml files is not the most user friendly. Sorry about that.

The way you usually know how to name files is to check the filename templates that are generated for each datatype.

The one with MPM should be right there:

https://bids-specification.readthedocs.io/en/latest/appendices/file-collections.html#magnetic-resonance-imaging

This will tell you for a given suffix what are the entities you can use, which ones are required (see the legend dropdown below the templates for more info on that) and in which order they should appear.

All filenames templates of all datatypes are also listed in a single page on the bids starter kit

https://bids-standard.github.io/bids-starter-kit/folders_and_files/files.html#anat-anatomical-mri

[CPernet]

done a PR just to move info in the main spec -- but I still think it makes no sense for MPM which entities are mandatory vs optional

• case 1, one acquires data using variable flip angles and one stores the various flips but not the inversion time (in the json a single value, makes sense)
• case 2, one acquires data using multiple inversion times and have the inversion time becomes inherently needed (while optional) to distinguish files but have to store the same flip many times (while being a single value, makes no sense)

[tsalo]

@CPernet the definition for "MPM" that we have in the specification is consistent with the entity requirements in the schema. If there is a canonical definition for the MPM protocol that differs from how we define it, and makes it clear what the requirements are, then can you link to that and we can update the definition?

[CPernet]

the entity requirement are 'correct' code wise, I am arguing here for reference that scientifically speaking this makes little sense to stores the various flips (mandatory) while the inversion time (optional) if using multiple flip angles data acquisition scheme (which is fine) but have to use the optional inversion time and store the same flip value many times when using a multiple inversion data acquisition scheme (ie either both mandatory or have some conditional if multiple flip, inversion is optional because always the same, if multiple inversion, flip is optional because always the same)