Implement time-synchronized import of motion and physio data

[sjeung]

As discussed with @arnodelorme and @dungscout96, here is the pull request that implements read-in of BIDS-formatted motion and physio data following the current specification (extension proposal still in motion's case).

@dungscout96 I still have to run some tests for multi-run and multi-session cases. You can already go ahead and review if you want (about the general structure of the code .etc), and I would also share some test data sets in BIDS format so you can properly try this.

[arnodelorme]

Thank you

[arnodelorme]

@dungscout96 the changes look ok to me, although you might want to look deeper at the changes related to channel location. @sjeung why did you modify the section importing channel location?

[sjeung]

@dungscout96 the changes look ok to me, although you might want to look deeper at the changes related to channel location. @sjeung why did you modify the section importing channel location?

Awesome, thanks Arno, I suggest the change because electrodes.tsv is optional in BIDS and I noticed that the chanloc part would throw an error when the file is not found. I also have to polish the metadata storage because currently almost nothing from motion or physio data is stored in bids struct in the script.

[arnodelorme]

That's strange. We have imported dozens of BIDS without electrodes.tsv - maybe it would throw an error when it does not find 10-20 electrodes? Do you have a dataset number so we can test?

[sjeung]

Sure, it's a super minimal data set with only one participant. Let me know if you were able to replicate the error, otherwise I will double check my setup. An older version from a year ago or so used to work without problem https://drive.google.com/drive/folders/1NCVe64hndfL7C2iAaswpBzRM1S7WtEDG?usp=sharing

[arnodelorme]

Thanks. Dung would you mind testing?

[dungscout96]

Thanks @sjeung for the code addition. Two general comments:

• Can you point to the BIDS motion spec from which this implementation follows? How are scans.tsv, motion.tsv, and physio.tsv supposed to be handled?
• In order to avoid monolithic code, it would be better if you can refactor line 562-760 into its own function. It seems to me that there're only few variables from before that the section depend on.

@arnodelorme This commit introduced the changes where electrodes.tsv and channels.tsv are expected to exist: https://github.com/sccn/bids-matlab-tools/commit/3b58ba60758f48f549e7b9bb37c417a38385747b#diff-d9ee0e01064e30c819fe9a2b79c08281ec47f91adf17b586b5e49d1b23af470d. This is why it's failing for Sein. Was there a reason for the change?

[dungscout96]

we fixed the error of not being able to import the BIDS dataset when no electrodes.tsv file is found in a58f356

[sjeung]

Thanks @sjeung for the code addition. Two general comments:

• Can you point to the BIDS motion spec from which this implementation follows? How are scans.tsv, motion.tsv, and physio.tsv supposed to be handled?
• In order to avoid monolithic code, it would be better if you can refactor line 562-760 into its own function. It seems to me that there're only few variables from before that the section depend on.

@arnodelorme This commit introduced the changes where electrodes.tsv and channels.tsv are expected to exist: 3b58ba6#diff-d9ee0e01064e30c819fe9a2b79c08281ec47f91adf17b586b5e49d1b23af470d. This is why it's failing for Sein. Was there a reason for the change?

Hi Dung, thanks for the input! Here is the current spec for motion data.

https://docs.google.com/document/d/1iaaLKgWjK5pcISD1MVxHKexB3PZWfE2aAC5HF_pCZWo/edit?usp=sharing

and yes, I will get to it and push the changes in the next days.

[sjeung]

Hello @dungscout96, I am going to work on this today and would like to check with you

1) whether you prefer it to be a subfunction inside pop_importbids.m or an extra mfile?

And

2) could you check whether you would then use this function for potentially more modalities (e.g., eyetracking)? Then I would name it something like "importbids_noneeg" or so (although MEG and iEEG are already covered in the main function).

[dungscout96]

Hi @sjeung, a subfunction as you are doing currently is good and make the extraction into external file easy if need to. I'd think this function will handle all modalities stored under */motion/ directories.