MEG BIDS specs - first draft

[Sylvain Baillet, Dr]

Hi everyone:

Following our discussion during the last telcon, we have prepared a draft as a first shot to proposing BIDS specs for MEG data: for now it is for “raw” data only (scalp sensor data that is, but can accommodate event-related and ongoing-recording designs): the actual “imaging" value of MEG is more in the (many) by-products of source analysis: we’ll probably get into that later down the road, once this first pass is well in place. 

It will readily extend to EEG and possibly, any time-series modality (e.g., NIRS and basic electrophysiology)  

The doc is accessible through the following link:

https://docs.google.com/document/d/1FWex_kSPWVh_f4rKgd5rxJmxlboAPtQlmBc1gyZlRZM/edit

We will be providing a full BIDS MEG dataset early next week (including anatomical MRI of the participant, which we realized today we’d rather deface, before public release, hence this slight delay).

Please let us know if you have comments and suggestions at this stage.

Cheers,

Sylvain.

                             

Sylvain Baillet, PhD

Professor, Neurology, Neurosurgery & Biomedical Engineering
Acting Director, McConnell Brain Imaging Centre

MNI Killam and FRQ-S Senior Scholar

Montreal Neurological Institute

McGill University 

http://mcgill.ca/bic

[PERNET Cyril]

Hi Sylvain

Excellent ! I'm CC in Arnaud as well as I'm sure he will have ideas for EEG. I'll have a look later on this week .. 

Cyril 

Dr Cyril Pernet

CCBS / Edinburgh imaging

Sent from my HTC

--
You received this message because you are subscribed to the Google Groups "The Brain Imaging Data Structure (BIDS) discussion" group.
To unsubscribe from this group and stop receiving emails from it, send an email to bids-discussi...@googlegroups.com.
To post to this group, send email to bids-di...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/bids-discussion/D5A2A29B-4323-47F6-BF41-5080920190B0%40mcgill.ca.
For more options, visit https://groups.google.com/d/optout.

[Sylvain Baillet, Dr]

Thank you, Cyril:

Yes: great if Arno can comment and contribute indeed.

All: the sample Public Domain dataset is now ready. Link is at the top of the MEG-BIDS document. 

Enjoy!   

Sylvain.

                             

Sylvain Baillet, PhD

Professor, Neurology, Neurosurgery & Biomedical Engineering
Acting Director, McConnell Brain Imaging Centre

MNI Killam and FRQ-S Senior Scholar

Montreal Neurological Institute

McGill University 

http://mcgill.ca/bic

To view this discussion on the web visit https://groups.google.com/d/msgid/bids-discussion/HE1PR05MB0986BE0431EFBA49A550ACCCBC8D0%40HE1PR05MB0986.eurprd05.prod.outlook.com.

For more options, visit https://groups.google.com/d/optout.

<Edinburgh University charitable status.txt>

[Arnaud Delorme]

This is interesting. I think it would also be worth to copy Christian Kothe and Nima Bigdely Shamlo and Tim Mullen who have developed the XDF format that could be compatible with what you are developing.

Arno

[Robert Oostenveld]

Hi Sylvain and others,

Thanks for the draft. I went over it this morning (and noticed that I am not the first one to start commenting). We have to consider how to do the governance on this document and how the decision process (working towards a final version) will be. There are some aspects that certainly require discussion, like how to deal with the specification of coordinates. Since it pertains to data sharing, the subject confidentiality and deidentification also requires thought.

At http://donders-institute.github.io/rdm-wiki/en/#!index.md you can find the documentation of the Research Data Management system that the Donders Institute is now finalizing. The system itself is not yet visible for outside users. The Donders RDM also covers data sharing, and I hope that BIDS can be adopted to maximize the value of the shared data for reuse. 

Seeing it in action with your shared tutorial dataset is very useful. I will also make similar datasets for some of our shared subejcts’ data (from http://bit.ly/22rUk4r), picking some variability in terms of data formats. I might also try to convert part of HCP MEG data in a (test, i.e. not with the actual data files) BIDS format.

best regards,

Robert

[Chris Filo Gorgolewski]

On Mon, Mar 21, 2016 at 6:51 AM, Robert Oostenveld <r.oost...@donders.ru.nl> wrote:

Hi Sylvain and others,

Thanks for the draft. I went over it this morning (and noticed that I am not the first one to start commenting). We have to consider how to do the governance on this document and how the decision process (working towards a final version) will be.

The way we approached it in BIDS Core was the following:

1. I would identify controversial issues from the discussion on the google doc and write an email to involved parties clearly explaining potential options and their pros and cons. Keeping those emails to the point was crucial. They were similar to those recently snt by William. 

2. We would discuss

3. In most cases a consensu was be reached.

 

There are some aspects that certainly require discussion, like how to deal with the specification of coordinates. Since it pertains to data sharing, the subject confidentiality and deidentification also requires thought.

Indeed anonimization should be addressed, but remain optional (BIDS will be also used within labs with no intention for sharing). I'm personally of the opinion that one should work on anonimized data even if there are no plans to share it (it's safer!), but in some cases this is not possible. For example some defacing algorithms used to remove identifiable features from T1 images badly interfere with source localization software used in EEG (at least so I was told).

At http://donders-institute.github.io/rdm-wiki/en/#!index.md you can find the documentation of the Research Data Management system that the Donders Institute is now finalizing. The system itself is not yet visible for outside users. The Donders RDM also covers data sharing, and I hope that BIDS can be adopted to maximize the value of the shared data for reuse.

This sounds great. Let me know if we should invite someone from Donders RDM to the mailing list.

Best,

Chris

[Robert Oostenveld]

Hi Chris

On 22 Mar 2016, at 04:17, Chris Filo Gorgolewski <krzysztof....@gmail.com> wrote:

The way we approached it in BIDS Core was the following:

1. I would identify controversial issues from the discussion on the google doc and write an email to involved parties clearly explaining potential options and their pros and cons. Keeping those emails to the point was crucial. They were similar to those recently snt by William. 

2. We would discuss

3. In most cases a consensu was be reached.

Sounds like a good approach to me.

At http://donders-institute.github.io/rdm-wiki/en/#!index.md you can find the documentation of the Research Data Management system that the Donders Institute is now finalizing. The system itself is not yet visible for outside users. The Donders RDM also covers data sharing, and I hope that BIDS can be adopted to maximize the value of the shared data for reuse.

This sounds great. Let me know if we should invite someone from Donders RDM to the mailing list.

I should have mentioned that I am myself member of the Donders RDM team, so I will bridge the two. I have already initiated discussions with some of the Donders RDM stakeholders regarding BIDS.

best

Robert

[Arnaud Delorme]

Dear all,

There does not seem to have been much progress on this lately. I would like to share some EEG data and use a version of this new format (I can always go back and fine tune my datasets later if it changes).

EEG manufacturer

What I see missing for EEG is mainly “EEG file format”. Is it OK to add this?

Events

About Events. Section 8.5.1 of the original BIDS specification describe event information for fMRI data (at the end of this message).

As far as event files are concerned, apparently there is no room to store the original event files. With EEG (and MEG), most often event files are collected separately from the EEG/MEG (text files from presentation etc…). Apparently the BIDS specification require that such files be recoded as tabular text (section 8.5.1) and the original event files not included - please correct me if I am wrong. This might be better because sometimes the first event(s) is(are) shifted (for example the first event in the EEG correspond to the second event in the event file or vice versa). Sometimes some events are missing from the EEG file (for example reaction time are only present in the event file). This can create issues when fusing the data. So in principle I would agree with recording such files although it is always good to have the original event files.

Electrode placement

The current BIDS document does not contain a consensus about electrode placement specification and coregistration. I personally agree with the tabular structure which is more intuitive than the JSON one. Could we reach a consensus on that?

Cheers,

Arno

The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.

[Chris Gorgolewski]

Since I am not leading the MEG/EEG effort I'll only comment on the events question and leave the other two to others,

On Tue, Jan 10, 2017 at 7:39 AM, Arnaud Delorme <ar...@ucsd.edu> wrote:

Dear all,

There does not seem to have been much progress on this lately. I would like to share some EEG data and use a version of this new format (I can always go back and fine tune my datasets later if it changes).

EEG manufacturer

What I see missing for EEG is mainly “EEG file format”. Is it OK to add this?

Events

About Events. Section 8.5.1 of the original BIDS specification describe event information for fMRI data (at the end of this message).

As far as event files are concerned, apparently there is no room to store the original event files. With EEG (and MEG), most often event files are collected separately from the EEG/MEG (text files from presentation etc…). Apparently the BIDS specification require that such files be recoded as tabular text (section 8.5.1) and the original event files not included - please correct me if I am wrong. This might be better because sometimes the first event(s) is(are) shifted (for example the first event in the EEG correspond to the second event in the event file or vice versa). Sometimes some events are missing from the EEG file (for example reaction time are only present in the event file). This can create issues when fusing the data. So in principle I would agree with recording such files although it is always good to have the original event files.

Indeed the compulsory format to store events in BIDS is a TSV file with onset and duration columns. This provides a normalized across presentation software way of storing data and makes measurement units and offsets unambiguous. The point of keeping the original logs (which sometimes are in proprietary formats) is well taken. There is a place for storing such logs - it's the sourcedata subdirectory (see section 3.4). However, the the source data files should be included only in addition to the TSV files - not as their replacement. The goal of BIDS is to present data in a normalized way that makes it easy for people to reuse and reanalyze.

I hope this helps!

Best,

Chris

To unsubscribe from this group and stop receiving emails from it, send an email to bids-discussion+unsubscribe@googlegroups.com.
To post to this group, send email to bids-discussion@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/bids-discussion/D5A2A29B-4323-47F6-BF41-5080920190B0%40mcgill.ca.
For more options, visit https://groups.google.com/d/optout.

The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.

--
You received this message because you are subscribed to the Google Groups "bids-discussion" group.
To unsubscribe from this group and stop receiving emails from it, send an email to bids-discussion+unsubscribe@googlegroups.com.
To post to this group, send email to bids-discussion@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/bids-discussion/46FF8ED3-2AE7-4618-AFB9-D2D118AC019E%40ucsd.edu.

[Robert Oostenveld]

Hi Arno,

For MEG the specification of the manufacturer also implies the (original) file format. For EEG that is indeed not the case.

There is no controlled vocabulary for EEG file formats, so the (machine readable/interpretable) description of the file format is not obvious. There are also subformats for some. Is it possible to deduce the file format from the file itself (extension or magic)? If so, then it would not be needed to add the file format to a sidecar file.

Perhaps more relevant is to review the EEG reference scheme, without which the data becomes more difficult (or more limited) to interpret. E.g. details such as CMS/DRL for Biosemi have not been considered yet.

best
Robert

> On 10 Jan 2017, at 16:39, Arnaud Delorme <ar...@ucsd.edu> wrote:
>
> Dear all,
>
> There does not seem to have been much progress on this lately. I would like to share some EEG data and use a version of this new format (I can always go back and fine tune my datasets later if it changes).
>
> EEG manufacturer
>
> What I see missing for EEG is mainly “EEG file format”. Is it OK to add this?
>
> Events
>
> About Events. Section 8.5.1 of the original BIDS specification describe event information for fMRI data (at the end of this message).
>
> As far as event files are concerned, apparently there is no room to store the original event files. With EEG (and MEG), most often event files are collected separately from the EEG/MEG (text files from presentation etc…). Apparently the BIDS specification require that such files be recoded as tabular text (section 8.5.1) and the original event files not included - please correct me if I am wrong. This might be better because sometimes the first event(s) is(are) shifted (for example the first event in the EEG correspond to the second event in the event file or vice versa). Sometimes some events are missing from the EEG file (for example reaction time are only present in the event file). This can create issues when fusing the data. So in principle I would agree with recording such files although it is always good to have the original event files.
>
> Electrode placement
>
> The current BIDS document does not contain a consensus about electrode placement specification and coregistration. I personally agree with the tabular structure which is more intuitive than the JSON one. Could we reach a consensus on that?
>
> Cheers,
>
> Arno
>

> <Screen Shot 2017-01-10 at 7.20.54 AM.png>

> You received this message because you are subscribed to the Google Groups "bids-discussion" group.

> To unsubscribe from this group and stop receiving emails from it, send an email to bids-discussi...@googlegroups.com.
> To post to this group, send email to bids-di...@googlegroups.com.

> To view this discussion on the web visit https://groups.google.com/d/msgid/bids-discussion/46FF8ED3-2AE7-4618-AFB9-D2D118AC019E%40ucsd.edu.