input parameters when dealing with a BIDS dataset

[Remi-Gau]

Just a quick head's up that might help with long term usage of the toolbox.

There is ongoing work to create a Brain imaging data structure extension proposal (BIDS BEP) to organize and curate MP2RAGE data.

The initial pull request on the BIDS repo is here and has since been broken down into several smaller pull-requests.
bids-standard/bids-specification#508

An example of what those MP2RAGE dataset would look like is available on OSF: https://osf.io/k4bs5/

The issue is that some of the terms will be different from those mentioned in the toolbox or those you find in DICOM header or on your MRI console.

I think the most notable are that:

• EchoSpacing I think becomes RepetitionTimeExcitation (though I am not entirely sure of that one because this number even though it "exists" on the MRI console, it appears nowhere in the DICOM header I have as far as I can tell).

• RepetionTime becomes RepetitionTimePreparation.

See an example of the metadata contained in a JSON file

{
"InversionTime": 0.8, 
"FlipAngle": 5, 
"RepetitionTimeExcitation": 0.0062, 
"RepetitionTimePreparation": 5.5,
 "NumberShots": 159, 
"FieldStrength":7
}

So to make sure that users are not confused, if they deal with this kind of data set it might be good to update the documentation (once the work on that BIDS BEP has been finalized).

[benoitberanger]

Hi @Remi-Gau,

Thank you for your message.

Assign ergonomic names to the parameters the function needs is tricky, since each manufacturer has different naming systems. This is also the case for dicom2nifti converters.

You mentioned a good example : what I called in this repo EchoSpacing comes from the Siemens name in the sequence parameter. This variable is not captured by dcm2niix JSON files (BIDS sidecar).

This is the reason I added in the description of this parameter a generic definition, which should be cross-manufacturer :

TR of the GRE readout

I also added the corresponding name on the scanner I have access :

On Siemens scanners, this is called EchoSpacing

I understand this solution is not optimal, and projects like BIDS tends to flatten the irregularities datasets by bypassing manufacturers and converters specificities.

If I understand correctly the purpose of the issue, I should add in the README and in the code the name defined by both dcm2niix and by BIDS specification ? I think it is a good idea.

Benoît

[Remi-Gau]

Thanks for the reply. @benoitberanger

Assign ergonomic names to the parameters the function needs is tricky

This is an understatement. 😉

This variable is not captured by dcm2niix JSON files (BIDS sidecar).

Ha that's typical the type of information that some users might need to know, so they don't waste time looking for it there.Do not hesitate to add it to the README. 🙏

I understand this solution is not optimal, and projects like BIDS tends to flatten the irregularities datasets by bypassing manufacturers and converters specificities.

If I understand correctly the purpose of the issue, I should add in the README and in the code the name defined by both dcm2niix and by BIDS specification ? I think it is a good idea.

Yes something along those lines could help. Sometimes I find that I don't want to reinvent the wheel when writing my README and adding a link to another page explains what I want to say can already be helpful.

That BIDS BEP for MP2RAGE is not yet merged but it is on a good path and I don't think that those terms will change (at least with BIDS version 1.X).

I can maybe ping you when it is official so you can update things. Let me know.

[benoitberanger]

I added a table in the README, to improve (I hope) readability for the parameters. 92074f9
Later, with the release of MP2RAGE as a native BIDS modality, I will add the corresponding fields. Please feel free to ping me when the new BIDS version is released.

[Remi-Gau]

The new BIDS version should be out soonish and the qMRI part that include MP2RAGE are now set and should not change anymore.

Really like the table you added in the README.

I think I will open an issue in the bids-starter kit to document some of those things for future bids users. Is it OK we end up reusing some of the content of that table into the starter kit?

I will most likely ping you and the authors of the BIPS extension for qMRI to make sure everyone is on the same page. 😉

[benoitberanger]

Please feel free to re-use the table that links the parameters. This might also be an opportunity to cross-check it !