Physiological Data - BIDS

Hi!

I am converting a multimodal dataset into the BIDS format, and I wanted to clarify some things. For context, I have all my eeg files in the eeg/ directories, and everything else (gaze, pupil size, head position, ecg, respiration, eog) in the beh/ directory as they are not neural recordings.

  1. As per the BIDS specifications, physiological recordings have the following template:
sub-<label>/[ses-<label>/]
    <datatype>/
        <matches>[_recording-<label>]_physio.tsv.gz
        <matches>[_recording-<label>]_physio.json

For the template directory name, <datatype> can correspond to any data recording modality, for example func, anat, dwi, meg, eeg, ieeg, or beh.

Is the datatype allowed to be only any of the above options? Or can I have ecg/, respiration/, eyetrack/ and so on as per my needs?

  1. I came across the BIDS Eyetracking extention proposal (BIDS Eye Tracking - Google Docs) in this forum. Can I use this specification for my eyetracking data or does the proposal have to be accepted for my dataset to pass the validator?

Thanks!

Hi @nikhilkuppa,

If the physio data were collected during a particular scan, they should go in that datatype’s folder. The particular kind of physio data (e.g., respiration) should be defined in the tsv column names and the accompanying JSON. For example, see how respiration is used in a fMRI experiment here: bids-examples/synthetic/sub-01/ses-01/func at 7de1e8c569ae70209217c42a6843be11ffc8e9fd · bids-standard/bids-examples · GitHub.

You can follow the proposal, but just have the files listed in a .bidsignore file in the root directory (e.g., sub*/ses*/*/*physio*) so they will not throw an error in the BIDS validator, and then remove that line when the proposal is integrated.

Best,
Steven

I see, these recordings were not collected during an imaging scan per se. They were collected from subjects while they watched different stimuli videos, and hence I had all the physio recordings in the beh/ directory and eeg recordings in the eeg/ directory for now.

My question is if I can add more datatypes than just func, anat, dwi, meg, eeg, ieeg, or beh, and if not, whether storing them in the beh/ directory is the right way to go?

I would say beh makes sense, in that case.