Corregistration between MEG and MRI

Guiomar Niso

unread,

Jan 23, 2018, 10:21:07 AM1/23/18

to bids-discussion, Sylvain Baillet, Dr, Vladimir Litvak, Rik Henson, François Tadel, Chris Filo Gorgolewski, Elizabeth Anne Bock, Ms, Alexandre Gramfort, Joseph Wexler, Robert Oostenveld, Mainak Jas

Dear all,

We are facing a challenge on the corregistration between MEG and MRI.

In the MEG-BIDS extension,

we have some fields to point to the corresponding MRI in the *_meg.json file:

Anatomical MRI:

[IntendedFor]:
[AnatomicalMRICoordinateSystem]:
[AnatomicalMRICoordinateUnits]:
[AnatomicalMRICoordinateSystemDescription]:

When we record MEG we also record some Landmarks, so they can be matched with the anatomical points in the MRI at the time of the corregistration.

We usually save these points in some kind of MEG Coordinates.

In the MEG-BIDS extension we store them in the *_meg.json file:

Anatomical landmarks:

[LandmarkCoordinates]:
[LandmarkCoordinateSystem]:
[LandmarkCoordinateUnits]:
[LandmarkCoordinateSystemDescription]:

However, these Landmark points have also an equivalent set(s) of Coordinates in MRI space.

Currently, there's no place to store these later.

So we are discussing about different options to solve this.

From our point of view these coordinates should be placed in the json file next to the corresponding MRI and Freesurfer independently, because if the MRI is modified, etc, there's no need to go to MEG json.

However, we are wondering if the MRI-BIDS could be modified to include this.

We are proposing to include the following fields in the MRI/Freesurfer, *.json files:

(8.3.1 Common metadata fields):

Anatomical landmarks (useful for co-registration):

[LandmarkCoordinates]: Key:value pairs of any number of additional anatomical landmarks and their coordinates relative to the associated anatomical MRI, interpreted following the ‘LandmarkCoordinateSystem’ (e.g. {"AC": [127,119,149], "PC": [128,93,141], "IH": [131,114,206]}, or {"NAS": [127,213,139], "LPA": [52,113,96], "RPA": [202,113,91]})
[LandmarkCoordinateSystem]: this refers to how the coordinates of the landmarks are to be interpreted (e.g. “DICOM”, “Analyze”, “Talairach-Tournoux”, “MNI”, “NIfTI”, “freesurfer”, “Paxinos-Franklin”, “Allen Institute”, etc). See appendix (https://docs.google.com/document/d/1FWex_kSPWVh_f4rKgd5rxJmxlboAPtQlmBc1gyZlRZM/edit#heading=h.snhwsmgj62e1)
[LandmarkCoordinateUnits]: Units in which the coordinates that are listed in the field ‘LandmarkCoordinateSystem’ are represented, e.g. “m”, “cm”, “mm”.
[LandmarkCoordinateSystemDescription]: Freeform description of the coordinate system. May also include a link to a documentation page or paper describing the system in greater detail.

Note that usually the original T1 MRI and the FreeSurfer MRI (in derivatives) do not have the same referential, so we need add them independently.

We were also considering the option of including a structure of nested fields:

{
'Structural1'
{
IntendedFor
LandmarkCoordinates
LandmarkCoordinateSystem
LandmarkCoordinateUnits
LandmarkCoordinateSystemDescription
}

'Structural2'
{
IntendedFor
LandmarkCoordinates
LandmarkCoordinateSystem
LandmarkCoordinateUnits
LandmarkCoordinateSystemDescription
}
....
}

Thus, the same set of points could be stored in multiple Coordinate systems.

What are your thoughts?

Thanks in advance!!

Cheers,

Julia Guiomar Niso Galán, PhD

Montreal Neurological Institute, McGill University

McConnell Brain Imaging Centre

guioma...@mcgill.ca

Chris Gorgolewski

unread,

Jan 28, 2018, 12:53:18 AM1/28/18

to Guiomar Niso, bids-discussion, Sylvain Baillet, Dr, Vladimir Litvak, Rik Henson, François Tadel, Elizabeth Anne Bock, Ms, Alexandre Gramfort, Joseph Wexler, Robert Oostenveld, Mainak Jas

Thanks for bringing this up and apologies for late reply. For simplicity, let me consider only the _T1w.json files (not FreeSurfer).

I think this addition makes sense, but we don't need (and should not include) LandmarkCoordinateSystem, LandmarkCoordinateUnits, LandmarkCoordinateSystemDescription since those would always be the same for nifti files.

The definition than would be:

LandmarkCoordinates - Key/value pairs of any number of named anatomical landmarks and their voxel coordinates relative to the associated anatomical scan. Keys MUST be strings. Values MUST be lists of numbers (potentially floats) in voxel units (NOT millimeters). LandmarkCoordinates can be useful for matching MRI and MEG scans. Example {"AC": [127,119,149], "PC": [128,93,141], "IH": [131,114,206]}, or {"NAS": [127,213,139], "LPA": [52,113,96], "RPA": [202,113,91]}.

Using voxel coordinates instead of scanner or "standard" space would allow us to avoid headaches with interpreting sform/qform.

WDYT?

Best,

Chris

JB Poline

unread,

Jan 28, 2018, 2:30:28 PM1/28/18

to The Brain Imaging Data Structure (BIDS) discussion, Guiomar Niso, Sylvain Baillet, Dr, Vladimir Litvak, Rik Henson, François Tadel, Elizabeth Anne Bock, Ms, Alexandre Gramfort, Joseph Wexler, Robert Oostenveld, Mainak Jas

Hi,

Just a quick thought: voxel unit means that the information will always have to be attached to the file, it may be that in some cases this information may need to be read independantly. As I am not sure we can escape the transformation headache in most cases, I wonder if mm wouldnt actually be better, since a nifti file will always give us the mm->vox transform.

cheers

JB

--
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/CAAQzouO7tN6Ur3K36YeX7EoNHU1PN4vTAffRfPabDeuDWrGK0g%40mail.gmail.com.

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

Chris Gorgolewski

unread,

Jan 28, 2018, 2:35:25 PM1/28/18

to bids-discussion, Guiomar Niso, Sylvain Baillet, Dr, Vladimir Litvak, Rik Henson, François Tadel, Elizabeth Anne Bock, Ms, Alexandre Gramfort, Joseph Wexler, Robert Oostenveld, Mainak Jas

Indeed the nifti header can give you mm->vox transformation, but depending on the code/sform/qform situation there could be more than one. It's just a can of worms that I think we should avoid opening.

Could give a practical example where reading the landmark coordinates without access to the file would be required/useful (and how not having access to mm->vox transformation - which is part of the header - would not be and issue)?

Best,

Chris

On Sun, Jan 28, 2018 at 11:30 AM, JB Poline <jbpo...@gmail.com> wrote:

Hi,

Just a quick thought: voxel unit means that the information will always have to be attached to the file, it may be that in some cases this information may need to be read independantly. As I am not sure we can escape the transformation headache in most cases, I wonder if mm wouldnt actually be better, since a nifti file will always give us the mm->vox transform.

cheers
JB

To view this discussion on the web visit https://groups.google.com/d/msgid/bids-discussion/CALMaa82pc%3D-JX6O3QM%2BjGuiGVre3CvL_TGEaWke9oSjKiVhPvA%40mail.gmail.com.

JB Poline

unread,

Jan 28, 2018, 3:30:34 PM1/28/18

to The Brain Imaging Data Structure (BIDS) discussion, Guiomar Niso, Sylvain Baillet, Dr, Vladimir Litvak, Rik Henson, François Tadel, Elizabeth Anne Bock, Ms, Alexandre Gramfort, Joseph Wexler, Robert Oostenveld, Mainak Jas

Sure, I understand that, the argument was that the can of worm will have to be opened anyway in most situations.

For instance, let's say you are mainly working with surface based analyses in MEG/EEG in the MNI space, you probably want to be able to grab the landmark coordinates without having to read the t1 nifti header, open the can of worms, and get the mm.

Honestly, I am not strongly advocating for this, just some thoughts. Having to keep the link to the file also has the advantage of a (weak) provenance component, so that's also in favor of the voxel solution. In a BIDS context that's reasonable.

cheers

JB

To view this discussion on the web visit https://groups.google.com/d/msgid/bids-discussion/CAAQzouNnCREV20FfN6iLQ8SMTjFXFCi5Mv66Kox%3DO1zPu__goQ%40mail.gmail.com.

Chris Gorgolewski

unread,

Jan 28, 2018, 7:49:17 PM1/28/18

to bids-discussion, Guiomar Niso, Sylvain Baillet, Dr, Vladimir Litvak, Rik Henson, François Tadel, Elizabeth Anne Bock, Ms, Alexandre Gramfort, Joseph Wexler, Robert Oostenveld, Mainak Jas

On Sun, Jan 28, 2018 at 12:30 PM, JB Poline <jbpo...@gmail.com> wrote:

Sure, I understand that, the argument was that the can of worm will have to be opened anyway in most situations.
For instance, let's say you are mainly working with surface based analyses in MEG/EEG in the MNI space, you probably want to be able to grab the landmark coordinates without having to read the t1 nifti header, open the can of worms, and get the mm.

Not sure how this relates to raw T1w images in subject space. Anyway if you work "in the MNI space" wouldn't all brains be coregistered and thus landmarks across all participants have the same coordinates? IT would be good if someone could flesh out the use of the LadmarkCoordinates for a nifti T1w files in context of MEG processing. I think I am missing something here.

Honestly, I am not strongly advocating for this, just some thoughts. Having to keep the link to the file also has the advantage of a (weak) provenance component, so that's also in favor of the voxel solution. In a BIDS context that's reasonable.

cheers
JB

To view this discussion on the web visit https://groups.google.com/d/msgid/bids-discussion/CALMaa80FJw%3D%3DOMFwMYPq1oG09ND9wFy9aiF8YhCWHgPeaucZNg%40mail.gmail.com.

Rik Henson

unread,

Jan 29, 2018, 3:41:00 AM1/29/18

to Chris Gorgolewski, bids-discussion, Guiomar Niso, Sylvain Baillet, Dr, Vladimir Litvak, François Tadel, Elizabeth Anne Bock, Ms, Alexandre Gramfort, Joseph Wexler, Robert Oostenveld, Mainak Jas

Usually, there are 3 anatomically-defined points (eg nasion and L+R pre-aricular) that are manually marked on the MRI, and digitised on the real head. The digitised points on the head are in the same space as other coils that can be detected by MEG, and therefore by minimising the distance between MRI and digitised landmarks, the MRI and MEG are coregistered. (Sometimes, superior coregistration is achieved by also digitising the scalp and fitting to the scalp surface automatically extracted from the MRI, which doesn’t need any landmarks for the MRI, but the fiducials are nearly always useful as well, if only for a measure of coregistration accuracy).

For this reason, this coregistration is nearly always done on the native MRI, rather than MNI-normalised.

I don’t have a strong view on whether the landmarks should be coded in mm or voxels (or both?)

Rik

---------------------------------------------------------------

Prof Richard Henson

Deputy Director
MRC Cognition and Brain Sciences Unit
University of Cambridge

15 Chaucer Road
Cambridge, CB2 7EF
England

EMAIL: rik.h...@mrc-cbu.cam.ac.uk
URL: http://www.mrc-cbu.cam.ac.uk/people/rik.henson/personal

TEL     +44 (0)1223 355 294 x501
FAX     +44 (0)1223 359 062
MOB     +44 (0)794 1377 345

---------------------------------------------------------------

Best,

Chris

o    [LandmarkCoordinates]: Key:value pairs of any number of additional anatomical landmarks and their coordinates relative to the associated anatomical MRI, interpreted following the ‘LandmarkCoordinateSystem’ (e.g. {"AC": [127,119,149], "PC": [128,93,141], "IH": [131,114,206]}, or {"NAS": [127,213,139], "LPA": [52,113,96], "RPA": [202,113,91]})

o    [LandmarkCoordinateSystem]: this refers to how the coordinates of the landmarks are to be interpreted (e.g. “DICOM”, “Analyze”, “Talairach-Tournoux”, “MNI”, “NIfTI”, “freesurfer”, “Paxinos-Franklin”, “Allen Institute”, etc). See appendix (https://docs.google.com/document/d/1FWex_kSPWVh_f4rKgd5rxJmxlboAPtQlmBc1gyZlRZM/edit#heading=h.snhwsmgj62e1)

o    [LandmarkCoordinateUnits]: Units in which the coordinates that are listed in the field ‘LandmarkCoordinateSystem’ are represented, e.g. “m”, “cm”, “mm”.

o    [LandmarkCoordinateSystemDescription]: Freeform description of the coordinate system. May also include a link to a documentation page or paper describing the system in greater detail.

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/CAAQzouO7tN6Ur3K36YeX7EoNHU1PN4vTAffRfPabDeuDWrGK0g%40mail.gmail.com.

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

--
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/CALMaa82pc%3D-JX6O3QM%2BjGuiGVre3CvL_TGEaWke9oSjKiVhPvA%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

--
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/CAAQzouNnCREV20FfN6iLQ8SMTjFXFCi5Mv66Kox%3DO1zPu__goQ%40mail.gmail.com.

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

--
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/CALMaa80FJw%3D%3DOMFwMYPq1oG09ND9wFy9aiF8YhCWHgPeaucZNg%40mail.gmail.com.

JB Poline

unread,

Jan 29, 2018, 8:52:20 AM1/29/18

to The Brain Imaging Data Structure (BIDS) discussion, Chris Gorgolewski, Guiomar Niso, Sylvain Baillet, Dr, Vladimir Litvak, François Tadel, Elizabeth Anne Bock, Ms, Alexandre Gramfort, Joseph Wexler, Robert Oostenveld, Mainak Jas

Hi,

if always attached to the T1, voxel is better !

cheers

JB

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/CAAQzouO7tN6Ur3K36YeX7EoNHU1PN4vTAffRfPabDeuDWrGK0g%40mail.gmail.com.

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

--
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/CALMaa82pc%3D-JX6O3QM%2BjGuiGVre3CvL_TGEaWke9oSjKiVhPvA%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

--
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/CAAQzouNnCREV20FfN6iLQ8SMTjFXFCi5Mv66Kox%3DO1zPu__goQ%40mail.gmail.com.

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

--
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/CALMaa80FJw%3D%3DOMFwMYPq1oG09ND9wFy9aiF8YhCWHgPeaucZNg%40mail.gmail.com.

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

--

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/3822A22C3FB6974099045A8DB62D39F901CCAF4033%40wsr-ex-101.mrc-cbsu.local.

Dora Hermes

unread,

Feb 8, 2018, 3:41:14 PM2/8/18

to bids-discussion

With iEEG, we have similar issues describing the coordinate system as MEG. The MEG specs now have a nice list of the coordinate systems description as in their Appendix. However, the BEP003 (common derivatives) now also has a table describing different spaces, that can be used as labels in _space-<label>.

https://docs.google.com/document/d/1Wwc4A6Mow4ZPPszDIWfCUCRNstn7d_zzaWPcfcHmgI4/edit?ts=5a7ca46a#heading=h.mqkmyp254xh6

How can we avoid confusion between these?

cheers,

Dora

Best,

Chris

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/CAAQzouO7tN6Ur3K36YeX7EoNHU1PN4vTAffRfPabDeuDWrGK0g%40mail.gmail.com.

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

--
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/CALMaa82pc%3D-JX6O3QM%2BjGuiGVre3CvL_TGEaWke9oSjKiVhPvA%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

--
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/CAAQzouNnCREV20FfN6iLQ8SMTjFXFCi5Mv66Kox%3DO1zPu__goQ%40mail.gmail.com.

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

--
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/CALMaa80FJw%3D%3DOMFwMYPq1oG09ND9wFy9aiF8YhCWHgPeaucZNg%40mail.gmail.com.

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

--
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.

Chris Gorgolewski

unread,

Feb 8, 2018, 5:59:23 PM2/8/18

to bids-discussion

I found the MEG coordinate system table quite confusing. I believe it has been copied from http://www.fieldtriptoolbox.org/faq/how_are_the_different_head_and_mri_coordinate_systems_defined. My understanding is that the intention of this was to be able to uniquely localize the position of MEG/EEG sensors in the physical space. However:

1) One of the coordinate system is "nifti" which provides little to no information about physical location. Each nifti file has a different origin and affine matrix making this system of little use (unless you are referring to a particular nifti file such as a T1w file, but then I would suggest renaming this system and clarifying the issue).

2) Similarly "freesurfer" coordinate system makes only sense if linked to a particular freesurfer conformed file - unless one meant fsaverage.

3) MNI coordinate system should be clarified more. In https://docs.google.com/document/d/1Wwc4A6Mow4ZPPszDIWfCUCRNstn7d_zzaWPcfcHmgI4/edit?ts=5a7ca46a#heading=h.mqkmyp254xh6 we specify a number of different MNI variants. I added this as a suggestion to the MEG spec.

4) Two of the listed coordinate systems are for mice - I think they got there by accident. I would leave them out for now and potentially bring them back in when we will get more feedback from animal experts.

5) Listing orientation is a little confusing, because it could mean orientation of voxel dimensions (how RAS/LAS is typically interpreted in context of MR) or physical dimension (how I believe this was ment to be interpreted).

6) I could not find any details of the brainstorm_mri coordinate system. This needs to be clarified more. It this a template?

7) I'm not sure why each coordinate system has assigned units when units are also specified in JSON files.

8) Finally I would suggest using camelcase naming to make this table more consistent with https://docs.google.com/document/d/1Wwc4A6Mow4ZPPszDIWfCUCRNstn7d_zzaWPcfcHmgI4/edit?ts=5a7ca46a#heading=h.mqkmyp254xh6

I hope this sets us on a path to more clarity.

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/8659b616-bb0b-4089-bb46-40f12b6f5f21%40googlegroups.com.

Sylvain Baillet, Dr

unread,

Feb 14, 2018, 11:19:37 AM2/14/18

to bids-discussion, Dora Hermes

Hi everyone:

Concerning MEG-BIDS, we came today to the conclusion that MEG-BIDS descriptors of an MRI coordinate system for cross-modality coregistration is not necessary.

The reason is that MEG coordinate systems are defined from anatomical fiducials. The detection of such fiducials in structural MRI is done with software. As such, their MRI coordinates are derivatives. MEG-BIDS, in its current version, is for the organization of raw data. Hence we concluded that all fields related to the specification of an MRI coordinate system shall be removed and fall under the broader scope of MRI derivatives.

Note to our EEG friends, incl. for simultaneous EEG-MEG data collection:

if EEG channel locations are digitized, their raw coordinates are expressed in the 3D digitizer’s coordinate system. Location of anatomical fiducials will be also in same coordinate system. Then re-alignement with MEG and/or MRI is straightforward, using software. We leave the option in MEG-BIDS for EEG companion channels to be located in same or different coordinate system: same is encouraged, but different remains possible as long as anatomical fiducials are documented in both.

Hope this clarifies.

Cheers,

Sylvain.

-

Sylvain Baillet, PhD

Professor, Montreal Neurological Institute

McGill University

> more

On Feb 14, 2018, at 7:05 AM, Guiomar Niso <guioma...@ctb.upm.es> wrote:

Hi!

I'm not sure my last message reached the bids-discussion list. Just in case :)

Cheers,

Julia Guiomar Niso Galán, PhD

Montreal Neurological Institute, McGill University
McConnell Brain Imaging Centre

guioma...@mcgill.ca

2018-02-13 12:53 GMT+01:00 Guiomar Niso <guioma...@ctb.upm.es>:

Hi Dora and Chris,

Thanks for the feedback!

1) One of the coordinate system is "nifti" which provides little to no information about physical location. Each nifti file has a different origin and affine matrix making this system of little use (unless you are referring to a particular nifti file such as a T1w file, but then I would suggest renaming this system and clarifying the issue).

In MEG we almost always refer to the T1w because I the file we mainly use for scource reconstructuon, but we accept any suggestion. We introduced this table to have an starting point. To be honest I don't have much experience with coordinate systems, so I let the experts advice :)

2) Similarly "freesurfer" coordinate system makes only sense if linked to a particular freesurfer conformed file - unless one meant fsaverage.

I'm not sure here, any opinion?

3) MNI coordinate system should be clarified more. In https://docs.google.com/document/d/1Wwc4A6Mow4ZPPszDIWfCUCRNstn7d_zzaWPcfcHmgI4/edit?ts=5a7ca46a#heading=h.mqkmyp254xh6 we specify a number of different MNI variants. I added this as a suggestion to the MEG spec.

Ok, we keep these suggestions.

4) Two of the listed coordinate systems are for mice - I think they got there by accident. I would leave them out for now and potentially bring them back in when we will get more feedback from animal experts.

Sounds good. Thanks!

5) Listing orientation is a little confusing, because it could mean orientation of voxel dimensions (how RAS/LAS is typically interpreted in context of MR) or physical dimension (how I believe this was ment to be interpreted).

Then we could just have a description field in the table (as it in in the BEP003).

6) I could not find any details of the brainstorm_mri coordinate system. This needs to be clarified more. It this a template?

We'll update this: more info can be found here: http://neuroimage.usc.edu/brainstorm/CoordinateSystems

7) I'm not sure why each coordinate system has assigned units when units are also specified in JSON files.

Ok. Let's remove it and just leave the description field in the table. I thought this units were generally associated with one particular type of coordinate system.

8) Finally I would suggest using camelcase naming to make this table more consistent with https://docs.google.com/document/d/1Wwc4A6Mow4ZPPszDIWfCUCRNstn7d_zzaWPcfcHmgI4/edit?ts=5a7ca46a#heading=h.mqkmyp254xh6

Ok for CamelCase. Should we merge the two tables together? To me it would make sense, so we have things unified.

Cheers,

To view this discussion on the web visit https://groups.google.com/d/msgid/bids-discussion/CAAQzouNPDs-n4yaKdqVW6Jq%2B4Mt_9R4%3DaMJoCsqikVY-VZNgtA%40mail.gmail.com.

Dora Hermes

unread,

Feb 14, 2018, 12:08:13 PM2/14/18

to Sylvain Baillet, Dr, bids-discussion

Hi Sylvain,

Thank you, this makes it clearer for iEEG. MEG has their own raw coordinates, whereas iEEG positions are always registered to an image, so we always use the space described in BEP003: common derivatives.

If we were to use these BEP003 spaces for iEEG (including stereo EEG), would that make for easy compatibility with Brainstorm? Say - if an iEEG coordinate system is described as T1w (which means millimeter coordinates corresponding to a T1 image), as far as I can see it looks like it would be straightforward to import these into Brainstorm, is that correct? Thanks!

cheers,

Dora

--

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Dora Hermes, PhD.

Postdoctoral Fellow, Brain Center Rudolf Magnus, UMC Utrecht

Visiting Postdoctoral Scholar, Stanford University

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Sylvain Baillet, Dr

unread,

Feb 14, 2018, 12:14:01 PM2/14/18

to bids-discussion, Dora Hermes

Absolutely: as long as the iEEG/SEEG electrode locations are attached to anatomical landmarks in a specific T1, they can be realigned to the MEG (in case of simultaneous recordings). I emphasize the need of anatomical landmarks, regardless of realignment with MEG actually, otherwise the T1 coordinate system is ill-defined.

Cheers,

Sylvain.

-

Sylvain Baillet, PhD

Professor, Montreal Neurological Institute

McGill University

> more

Vladimir Litvak

unread,

Feb 15, 2018, 7:12:17 AM2/15/18

to bids-discussion, Rik Henson

Hi Sylvain,

How do you suggest that we deal with the practical problem of including coregistration information from Rik in the BIDS version of his dataset? This is quite a pressing issue for us as resolving it is the last thing keeping us from submitting our SPM special topic paper.

Vladimir

To view this discussion on the web visit https://groups.google.com/d/msgid/bids-discussion/FBBA4C6C-F2F3-4821-9D96-EA52099738F0%40mcgill.ca.

Rik Henson

unread,

Feb 15, 2018, 12:05:06 PM2/15/18

to Vladimir Litvak, bids-discussion, Sylvain Baillet

Sylvain – the logic makes sense – but are the MRI BIDS experts happy to include a new *.json file coding the fiducial coordinates (or any anatomical landmarks) associated with an MRI image? The last discussion we had on this a few weeks ago concerned whether those coordinates refer to voxel space or world space; information that could presumably easily be included in the *.json. I attach an example – can we agree on this? We need to agree soon to make some progress.

Best,

Chris

To view this discussion on the web visit https://groups.google.com/d/msgid/bids-discussion/8659b616-bb0b-4089-bb46-40f12b6f5f21%40googlegroups.com.

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

--
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/CAAQzouNPDs-n4yaKdqVW6Jq%2B4Mt_9R4%3DaMJoCsqikVY-VZNgtA%40mail.gmail.com.

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

--
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/FBBA4C6C-F2F3-4821-9D96-EA52099738F0%40mcgill.ca.

sub-01_ses-mri_acq-mprage_T1w_fid.json

Chris Gorgolewski

unread,

Feb 15, 2018, 12:11:54 PM2/15/18

to bids-discussion, Vladimir Litvak, Sylvain Baillet

The way I understood Julia the proposal was to add fiducial related fields to _T1w.json files (instead of creating new ones). So for your example the

sub-01/ses-mri/anat/sub-01_ses-mri_acq-mprage_T1w.json file would have one extra field:

"AnatomicalLandmarkCoordinates": {
  "Nasion": [4.3,119.2,8.300000000000001],
  "LPA": [75.3,24.6,31.9],
  "RPA": [81.2,20.6,33.5]
 }

(with the note, which is my proposal, that coordinates are in voxel usints to avoid having to deal with sform/qform confusion)/

Best,

Chris

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/8659b616-bb0b-4089-bb46-40f12b6f5f21%40googlegroups.com.

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

--
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/CAAQzouNPDs-n4yaKdqVW6Jq%2B4Mt_9R4%3DaMJoCsqikVY-VZNgtA%40mail.gmail.com.

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

--
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/FBBA4C6C-F2F3-4821-9D96-EA52099738F0%40mcgill.ca.

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

--

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/3822A22C3FB6974099045A8DB62D39F901D9F4C1A7%40wsr-ex-101.mrc-cbsu.local.

Guiomar Niso

unread,

Feb 15, 2018, 9:22:22 PM2/15/18

to bids-discussion, Chris Filo Gorgolewski, Dora Hermes, Sylvain Baillet, Dr

Hi!

I'm not sure my last message reached the bids-discussion list. Just in case :)

Cheers,

Julia Guiomar Niso Galán, PhD

Montreal Neurological Institute, McGill University

McConnell Brain Imaging Centre

guioma...@mcgill.ca

2018-02-13 12:53 GMT+01:00 Guiomar Niso <guioma...@ctb.upm.es>:

Hi Dora and Chris,

Thanks for the feedback!

1) One of the coordinate system is "nifti" which provides little to no information about physical location. Each nifti file has a different origin and affine matrix making this system of little use (unless you are referring to a particular nifti file such as a T1w file, but then I would suggest renaming this system and clarifying the issue).

In MEG we almost always refer to the T1w because I the file we mainly use for scource reconstructuon, but we accept any suggestion. We introduced this table to have an starting point. To be honest I don't have much experience with coordinate systems, so I let the experts advice :)

2) Similarly "freesurfer" coordinate system makes only sense if linked to a particular freesurfer conformed file - unless one meant fsaverage.

I'm not sure here, any opinion?

3) MNI coordinate system should be clarified more. In https://docs.google.com/document/d/1Wwc4A6Mow4ZPPszDIWfCUCRNstn7d_zzaWPcfcHmgI4/edit?ts=5a7ca46a#heading=h.mqkmyp254xh6 we specify a number of different MNI variants. I added this as a suggestion to the MEG spec.

Ok, we keep these suggestions.

4) Two of the listed coordinate systems are for mice - I think they got there by accident. I would leave them out for now and potentially bring them back in when we will get more feedback from animal experts.

Sounds good. Thanks!

5) Listing orientation is a little confusing, because it could mean orientation of voxel dimensions (how RAS/LAS is typically interpreted in context of MR) or physical dimension (how I believe this was ment to be interpreted).

Then we could just have a description field in the table (as it in in the BEP003).

6) I could not find any details of the brainstorm_mri coordinate system. This needs to be clarified more. It this a template?

We'll update this: more info can be found here: http://neuroimage.usc.edu/brainstorm/CoordinateSystems

7) I'm not sure why each coordinate system has assigned units when units are also specified in JSON files.

Ok. Let's remove it and just leave the description field in the table. I thought this units were generally associated with one particular type of coordinate system.

8) Finally I would suggest using camelcase naming to make this table more consistent with https://docs.google.com/document/d/1Wwc4A6Mow4ZPPszDIWfCUCRNstn7d_zzaWPcfcHmgI4/edit?ts=5a7ca46a#heading=h.mqkmyp254xh6

Ok for CamelCase. Should we merge the two tables together? To me it would make sense, so we have things unified.

Cheers,

Julia Guiomar Niso Galán, PhD

Montreal Neurological Institute, McGill University
McConnell Brain Imaging Centre
guioma...@mcgill.ca

2018-02-08 23:58 GMT+01:00 Chris Gorgolewski <krzysztof.gorgolewski@gmail.com>:

To view this discussion on the web visit https://groups.google.com/d/msgid/bids-discussion/CAAQzouNPDs-n4yaKdqVW6Jq%2B4Mt_9R4%3DaMJoCsqikVY-VZNgtA%40mail.gmail.com.

Guiomar Niso

unread,

Feb 15, 2018, 9:22:22 PM2/15/18

to bids-discussion

Hi Dora and Chris,

Thanks for the feedback!

1) One of the coordinate system is "nifti" which provides little to no information about physical location. Each nifti file has a different origin and affine matrix making this system of little use (unless you are referring to a particular nifti file such as a T1w file, but then I would suggest renaming this system and clarifying the issue).

In MEG we almost always refer to the T1w because I the file we mainly use for scource reconstructuon, but we accept any suggestion. We introduced this table to have an starting point. To be honest I don't have much experience with coordinate systems, so I let the experts advice :)

2) Similarly "freesurfer" coordinate system makes only sense if linked to a particular freesurfer conformed file - unless one meant fsaverage.

I'm not sure here, any opinion?

3) MNI coordinate system should be clarified more. In https://docs.google.com/document/d/1Wwc4A6Mow4ZPPszDIWfCUCRNstn7d_zzaWPcfcHmgI4/edit?ts=5a7ca46a#heading=h.mqkmyp254xh6 we specify a number of different MNI variants. I added this as a suggestion to the MEG spec.

Ok, we keep these suggestions.

4) Two of the listed coordinate systems are for mice - I think they got there by accident. I would leave them out for now and potentially bring them back in when we will get more feedback from animal experts.

Sounds good. Thanks!

5) Listing orientation is a little confusing, because it could mean orientation of voxel dimensions (how RAS/LAS is typically interpreted in context of MR) or physical dimension (how I believe this was ment to be interpreted).

Then we could just have a description field in the table (as it in in the BEP003).

6) I could not find any details of the brainstorm_mri coordinate system. This needs to be clarified more. It this a template?

We'll update this: more info can be found here: http://neuroimage.usc.edu/brainstorm/CoordinateSystems

7) I'm not sure why each coordinate system has assigned units when units are also specified in JSON files.

Ok. Let's remove it and just leave the description field in the table. I thought this units were generally associated with one particular type of coordinate system.

8) Finally I would suggest using camelcase naming to make this table more consistent with https://docs.google.com/document/d/1Wwc4A6Mow4ZPPszDIWfCUCRNstn7d_zzaWPcfcHmgI4/edit?ts=5a7ca46a#heading=h.mqkmyp254xh6

Ok for CamelCase. Should we merge the two tables together? To me it would make sense, so we have things unified.

Cheers,

Julia Guiomar Niso Galán, PhD

Montreal Neurological Institute, McGill University

McConnell Brain Imaging Centre

guioma...@mcgill.ca

To view this discussion on the web visit https://groups.google.com/d/msgid/bids-discussion/CAAQzouNPDs-n4yaKdqVW6Jq%2B4Mt_9R4%3DaMJoCsqikVY-VZNgtA%40mail.gmail.com.

Sylvain Baillet, Dr

unread,

Feb 16, 2018, 12:35:20 PM2/16/18

to Chris Gorgolewski, Vladimir Litvak, Rik Henson, bids-discussion

Chris, Rik and Vlad: thanks

I think it is preferable to simplify as much as possible the MEG-BIDS 1.0 implementation of these tricky inter-modal coordinate system issues - for the following reasons:

Let's consider the coordinates of anatomical landmarks in the T1 MRI volume as a derivatives: they are the produce of co-registration, one of the first processing steps of MEG/MRI raw data combination. That’s why I suggest we treat them derivatives, and therefore we shall not worry about providing their MRI coordinates for MEG-BIDS 1.0 or BIDS 1.1. Hence we will not use and extra field for sub-01/ses-mri/anat/sub-01_ses-mri_acq-mprage_T1w.json.

The MEG *_coordsystem.json fields offer very flexible ways of documenting coordinates and defining a coordinate system for scalp points, EEG channels, landmarks, MEG head coils etc.

I think the reality is that the coordinates of MEG channels will always be in the manufacturer’s head coordinate system.

However, it is possible that EEG, scalp, landmarks and head localization coils be expressed in the 3-D digitizer’s arbitrary coordinate system. Hence we need a field that explains this and I think we need to make mandatory that all these 3D points be expressed in the same coordinate system, for consistency and registration with MEG. What do you think?

In principle you could provide the landmarks in MRI coordinates in that *_coordsystem.json file (attached to the MEG, not the MRI) as long as you explain how the system is defined, what the units are, etc. AND that electrodes and other ancillary 3D points are in that same CS.

My recommendation would be though that you do not bother converting all these points to a MRI CS, and rather document them in any 3D CS. The registration with MRI will be done with your preferred analysis software.

Makes senses?

(Sorry for the long note)

Sylvain.

-

Sylvain Baillet, PhD

Professor, Montreal Neurological Institute

McGill University

> more

Rik Henson

unread,

Feb 16, 2018, 12:44:56 PM2/16/18

to Sylvain Baillet, Dr, Chris Gorgolewski, Vladimir Litvak, bids-discussion

No, sorry, I am more confused now than I was before! ;-)

The MRI coordinates of the fiducials are essential to provide with any MEG dataset, even if regarded as “derived”. Otherwise, MEG data cannot be localised. So in what JSON should we put them please?

R

---------------------------------------------------------------

Professor Richard Henson

Deputy Director
MRC Cognition and Brain Sciences Unit
University of Cambridge

15 Chaucer Road
Cambridge, CB2 7EF
England

EMAIL: rik.h...@mrc-cbu.cam.ac.uk
URL: http://www.mrc-cbu.cam.ac.uk/people/rik.henson/personal

TEL     +44 (0)1223 355 294 x501
FAX     +44 (0)1223 359 062
MOB     +44 (0)794 1377 345

---------------------------------------------------------------

From: Sylvain Baillet, Dr [mailto:sylvain...@mcgill.ca]
Sent: 16 February 2018 17:35
To: Chris Gorgolewski <krzysztof....@gmail.com>; Vladimir Litvak <v.li...@ucl.ac.uk>; Rik Henson <Rik.H...@mrc-cbu.cam.ac.uk>
Cc: bids-discussion <bids-di...@googlegroups.com>
Subject: Re: [bids-discussion] Corregistration between MEG and MRI

Chris, Rik and Vlad: thanks

Rik Henson

unread,

Feb 16, 2018, 12:50:44 PM2/16/18

to Sylvain Baillet, Dr, Vladimir Litvak, bids-discussion, Chris Gorgolewski (krzysztof.gorgolewski@gmail.com)

PS I’m not sure what you mean by “produce of coregistration”? They are manually defined by the researcher, not produced by any automated coregistration algorithm.

PPS MEG-BIDS does allow a JPG to help a user see where the fiducials were on the head, but this is unlikely to be accurate enough for the user to reproduce the skills of the researcher in finding them on the MRI (and the user may not have software to help manually define on the MRI).

---------------------------------------------------------------

Professor Richard Henson

Deputy Director
MRC Cognition and Brain Sciences Unit
University of Cambridge

15 Chaucer Road
Cambridge, CB2 7EF
England

EMAIL: rik.h...@mrc-cbu.cam.ac.uk
URL: http://www.mrc-cbu.cam.ac.uk/people/rik.henson/personal

TEL     +44 (0)1223 355 294 x501
FAX     +44 (0)1223 359 062
MOB     +44 (0)794 1377 345

---------------------------------------------------------------

From: Sylvain Baillet, Dr [mailto:sylvain...@mcgill.ca]

Sent: 16 February 2018 17:35

To: Chris Gorgolewski <krzysztof....@gmail.com>; Vladimir Litvak <v.li...@ucl.ac.uk>; Rik Henson <Rik.H...@mrc-cbu.cam.ac.uk>
Cc: bids-discussion <bids-di...@googlegroups.com>
Subject: Re: [bids-discussion] Corregistration between MEG and MRI

Chris, Rik and Vlad: thanks

Sylvain Baillet, Dr

unread,

Feb 16, 2018, 1:14:36 PM2/16/18

to Rik Henson, Vladimir Litvak, bids-discussion, Chris Gorgolewski (krzysztof.gorgolewski@gmail.com)

Thanks Rik:

PS I’m not sure what you mean by “produce of coregistration”? They are manually defined by the researcher, not produced by any automated coregistration algorithm.

Correct: that’s why I think we may consider their MRI coordinates as the result of a processing step (manual or automatic realignment, both using software - note that the MEG-MRI registration is meant for source imaging, for which software that does registration is required)

PPS MEG-BIDS does allow a JPG to help a user see where the fiducials were on the head, but this is unlikely to be accurate enough for the user to reproduce the skills of the researcher in finding them on the MRI (and the user may not have software to help manually define on the MRI).

I was thinking the user could localize these fiducials in the MRI themselves - indeed the photos help, and agreed this would not be perfect. But I would be provocative by saying that even expert MRI localization would not be perfect.

Best approach would be to use the EEG electrodes or extra scalp points digitized at the session to automatically refine the realignment b/w MRI and MEG with software (Brainstorm and probably others do that).

Alternative would be to have another optional field in *_coordsystem.json for landmark coordinates (and only landmarks) expressed in an MRI coordinate system (but then we’re stuck again with the pb of defining an MRI coordinate system, which is I think something MRI BIDS should come up with so that we follow their recommendations).

Does it make sense?

Sylvain.

Rik Henson

unread,

Feb 16, 2018, 1:26:53 PM2/16/18

to Sylvain Baillet, Dr, Vladimir Litvak, bids-discussion, Chris Gorgolewski (krzysztof.gorgolewski@gmail.com)

PS I’m not sure what you mean by “produce of coregistration”? They are manually defined by the researcher, not produced by any automated coregistration algorithm.

Correct: that’s why I think we may consider their MRI coordinates as the result of a processing step (manual or automatic realignment, both using software - note that the MEG-MRI registration is meant for source imaging, for which software that does registration is required)

Well, this becomes semantics – you could regard the MRI fiducial coordinates as raw data acquired by the researcher, or you could regard them as derived by the researcher at some later point – I don’t really mind, but…

PPS MEG-BIDS does allow a JPG to help a user see where the fiducials were on the head, but this is unlikely to be accurate enough for the user to reproduce the skills of the researcher in finding them on the MRI (and the user may not have software to help manually define on the MRI).

I was thinking the user could localize these fiducials in the MRI themselves - indeed the photos help, and agreed this would not be perfect. But I would be provocative by saying that even expert MRI localization would not be perfect.

Best approach would be to use the EEG electrodes or extra scalp points digitized at the session to automatically refine the realignment b/w MRI and MEG with software (Brainstorm and probably others do that).

….yes, most software packages use digitised headshape points (and EEG electrode locations when present) to improve coregistration. However, it is helpful to also have some fixed *anatomical* landmarks, eg to check on quality of coregistration by headshape (assuming minimal error in the manual definition of fiducials on MRI). For example, if the headshape points do not include nose and other non-smooth features, we often find that cranial headshape points can “slip” quite a lot over the “hemispherical” scalp, ie fit equally well after minor rotations.

Alternative would be to have another optional field in *_coordsystem.json for landmark coordinates (and only landmarks) expressed in an MRI coordinate system (but then we’re stuck again with the pb of defining an MRI coordinate system, which is I think something MRI BIDS should come up with so that we follow their recommendations).

I still think they are best in a JSON attached to the MRI file, because they are a property of the MRI, not the MEG (and to avoid mixing coordinate systems in the MEG JSON), but I don’t mind if it is easier to put in MEG JSON for the moment. However, we do need them somewhere in the example BIDS datasets, otherwise we’ll have one or more example analysis papers in the Frontiers Special Issue that get the fiducials from some other website, which is not ideal (since would illustrate limitations of MEG BIDS)?

Vladimir Litvak

unread,

Feb 16, 2018, 1:44:58 PM2/16/18

to Rik Henson, Sylvain Baillet, Dr, bids-discussion, Chris Gorgolewski (krzysztof.gorgolewski@gmail.com)

Hi all,

I’m with Rik on this one. I understand the considerations of elegance and purity of the standard but it’s just very strange to not provide any means of sharing this information which is one of the basic pieces of data for many analyses. Yes it is derived but it’s also the kind of thing that is important to share to reproduce one’s results.

Vladimir

Sylvain Baillet, Dr

unread,

Feb 19, 2018, 10:48:29 AM2/19/18

to Vladimir Litvak, Rik Henson, bids-discussion, Chris Gorgolewski (krzysztof.gorgolewski@gmail.com)

Thanks for the useful feedback, folks.

I agree with your suggestion, so let’s stick to the original proposition from Guio, reminded by Chris below:

Add one recommended field to the T1w json: no new file and no extra field in the MEG *_coordsystem.json needed.

sub-01/ses-mri/anat/sub-01_ses-mri_acq-mprage_T1w.json

"AnatomicalLandmarkCoordinates": {

  "Nasion": [134,12,34],

  "LPA": [37,43,35],

  "RPA": [170, 67,33]

As Chris suggested: the coordinates will be in voxels (ie, integers).

Is it fair to say we don’t define the coordinate system used for these points in the T1? It should be relatively straightforward to figure out which one was used from visualizing the T1. The reason I am asking is that it looks like the definition of CS is not finalized in MRI-BIDS. We could also add a field in _T1w.json for a freeform description of the CS e.g., “X=Anterior, Y=Left, Z=Dorsal”.

Please let me know how it sounds.

Thanks again,

Sylvain.

-

Sylvain Baillet, PhD

Professor, Montreal Neurological Institute

McGill University

> more

Rik Henson

unread,

Feb 20, 2018, 8:57:20 AM2/20/18

to Sylvain Baillet, Dr, Vladimir Litvak, bids-discussion, Chris Gorgolewski (krzysztof.gorgolewski@gmail.com)

Sounds great to me – if I create these new JSONs for T1s (with voxel indices), will it be uploadable to OpenNeuro – ie pass BIDS validators?

---------------------------------------------------------------

Prof Richard Henson

Sylvain Baillet, Dr

unread,

Feb 20, 2018, 9:19:07 AM2/20/18

to Rik Henson, Vladimir Litvak, bids-discussion, Chris Gorgolewski (krzysztof.gorgolewski@gmail.com), Mainak Jas

Hi Rik:

Please note that these are not new json files: the fiducial information will be indicated in a new field of the T1w.json (e.g., ub-01/ses-mri/anat/sub-01_ses-mri_acq-mprage_T1w.json) as shown in the example below.

I will finalize the MEG-BIDS specs doc and add this to the main BIDS spec doc today.

Hope this clarifies,

Sylvain.

-

Sylvain Baillet, PhD

Professor, Montreal Neurological Institute

McGill University

> more

Rik Henson

unread,

Feb 20, 2018, 9:30:55 AM2/20/18

to Sylvain Baillet, Dr, Vladimir Litvak, bids-discussion, Chris Gorgolewski (krzysztof.gorgolewski@gmail.com), Mainak Jas

Ok – then that is a coincidence then, because we did not have any JSON files for T1s in the ds000117 dataset – so I created ones called:

sub-01/ses-mri/anat/sub-01_ses-mri_acq-mprage_T1w.json

I must have just been lucky enough to guess the correct BIDS name! Are there any other fields (metadata) that need to go into that JSON file (apart from fiducials) – I cannot find anything obligatory from skimming http://bids.neuroimaging.io/bids_spec1.0.0.pdf?

Anyway, I’ll create all these for the ds000117 dataset, but will not be able to upload to OpenNeuro until the MEG part of the BIDS validator has been updated (I sent full list of errors to Joe in a previous email).

R

Sylvain Baillet, Dr

unread,

Feb 20, 2018, 10:13:39 AM2/20/18

to Rik Henson, Vladimir Litvak, bids-discussion, Chris Gorgolewski (krzysztof.gorgolewski@gmail.com), Mainak Jas, Guiomar Niso

Good.

It seems all the fields in there are recommended. Please refer to current working version of the MR document:

https://docs.google.com/document/d/1HFUkAEE-pB-angVcYe6pf_-fVf4sCpOHKesUvfb8Grc/edit?usp=sharing

and search for “AnatomicalLandmarkCoordinates"

as for the validator, I understand Mainak will finalize after a competing deadline on his end.

Guio: please double check the above and let us know the latest about the validator: thanks!

Sylvain.

-

Sylvain Baillet, Dr

unread,

Feb 20, 2018, 6:11:14 PM2/20/18

to bids-di...@googlegroups.com, Rik Henson, Vladimir Litvak, Chris Gorgolewski (krzysztof.gorgolewski@gmail.com), Mainak Jas, Guiomar Niso, Elizabeth Anne Bock, Ms, Francois Tadel, Mr, Martin Pierre Cousineau, M

Folks:

Here is the final version of the MEG-BIDS specs:

https://docs.google.com/document/d/1FWex_kSPWVh_f4rKgd5rxJmxlboAPtQlmBc1gyZlRZM/edit?usp=sharing

Anatomical landmarks are RECOMMENDED in the “AnatomicalLandmarkCoordinates” field of https://docs.google.com/document/d/1HFUkAEE-pB-angVcYe6pf_-fVf4sCpOHKesUvfb8Grc/edit?usp=sharing

Mainak: please let us know when the validator is updated (thanks!) so that we can post the final MEG sample datasets and close the submission of the manuscript.

Thanks all!

Sylvain.

-

Sylvain Baillet, PhD

Professor, Montreal Neurological Institute

McGill University

> more

--
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/3BC789E6-F43D-418B-96B3-D9BBA7AF2C69%40mcgill.ca.

Satrajit Ghosh

unread,

Feb 20, 2018, 6:22:12 PM2/20/18

to bids-di...@googlegroups.com, Rik Henson, Vladimir Litvak, Mainak Jas, Guiomar Niso, Elizabeth Anne Bock, Ms, Francois Tadel, Mr, Martin Pierre Cousineau, M

hi sylvain,

i left a comment in the anatomical landmark coordinate section to deal with multiple MEG sessions. i think the last entry in the comment clarifies my position. i'll paste a modified version here for the group.

---

scenario: single MRI session, multiple MEG sessions for a participant

it seems that you will need to add a new set of coordinates for every new MEG session for a given person in the T1w json file. and this would have to refer back to the session in question.

aren't the anatomical landmarks derived from the MEG session fiducial + digitization info rather than being prescribed in anatomical space? i.e. there is going to be variability of where the fiducials are placed during each MEG session.

to me it would seem, that these anatomical landmarks {"AC": [127,119,149], "PC": [128,93,141], "IH": [131,114,206]}, or {"NAS": [127,213,139], "LPA": [52,113,96], "RPA": [202,113,91]} would be derived from each MEG session by mapping on to the T1, and hence you would have different coordinates for each session.

as such it seems the best place for such coordinates is alongside the MEG sidecar json rather than the t1w sidecar json

---

if this is a non-issue, please feel free to resolve the comment, perhaps with an explanation of how multiple session fiducials are handled.

cheers,

satra

Sylvain.

-

EMAIL: rik.henson@mrc-cbu.cam.ac.uk
URL: http://www.mrc-cbu.cam.ac.uk/people/rik.henson/personal

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/3BC789E6-F43D-418B-96B3-D9BBA7AF2C69%40mcgill.ca.
For more options, visit https://groups.google.com/d/optout.

--
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/4B972926-3FB1-49FF-A98F-8BBCC119063F%40mcgill.ca.

Chris Gorgolewski

unread,

Feb 20, 2018, 6:30:05 PM2/20/18

to bids-discussion, Rik Henson, Vladimir Litvak, Mainak Jas, Guiomar Niso, Elizabeth Anne Bock, Ms, Francois Tadel, Mr, Martin Pierre Cousineau, M

In addition and unrelated to Satra's comments I had a couple of thoughts:

1) Currently the field is called AnatomicalLandmarkCoordinates. Is the qualifier "anatomical" likely to be accurate for all use cases? Perhaps using a more general name "LandmarkCoordinates" would be more future proof.

2) I know that originally I argue for using voxels as coordinate units (to avoid having to deal with qform/sform), but if mm are more desirable (for example to be consistent with potential future use of landmarks described in the space of a template such as one ofthe MNI templates) one can just specify in the definition that only the qform will be used to interpret the "mm" coordinates as voxel locations.

Apologies for late comments - I want to avoid potential future regrets :)

Best,

Chris

cheers,

satra

Sylvain.

-

EMAIL: rik.h...@mrc-cbu.cam.ac.uk
URL: http://www.mrc-cbu.cam.ac.uk/people/rik.henson/personal

To view this discussion on the web visit https://groups.google.com/d/msgid/bids-discussion/CA%2BA4wOmfgdRP7qQOSMbV4OKCuen7GA-LvwB_iQFY-pqfsO7jvg%40mail.gmail.com.

Rik Henson

unread,

Feb 21, 2018, 2:46:23 AM2/21/18

to Satrajit Ghosh, bids-di...@googlegroups.com, Vladimir Litvak, Mainak Jas, Guiomar Niso, Elizabeth Anne Bock, Ms, Francois Tadel, Mr, Martin Pierre Cousineau, M

Sorry – I don’t understand why you would need new anatomical landmarks for each MEG session in the MRI JSON (apologies if I misunderstood Satra’s suggestion below). These landmarks are a property of the MRI image, not of the MEG data. They are *before* any coregistration of the MEG and MRI. (Note that their coordinates are likely to differ from the same landmarks that are digitized in the MEG JSON, for which there could be one JSON per MEG session to allow for different digitizations in different sessions). So I cannot see why there would need to be more than one set in the MRI JSON?

R

---------------------------------------------------------------

Professor Richard Henson

Deputy Director
MRC Cognition and Brain Sciences Unit
University of Cambridge

15 Chaucer Road
Cambridge, CB2 7EF
England

EMAIL: rik.h...@mrc-cbu.cam.ac.uk
URL: http://www.mrc-cbu.cam.ac.uk/people/rik.henson/personal

TEL     +44 (0)1223 355 294 x501
FAX     +44 (0)1223 359 062
MOB     +44 (0)794 1377 345

---------------------------------------------------------------

From: Satrajit Ghosh [mailto:sa...@mit.edu]
Sent: 20 February 2018 23:22
To: bids-di...@googlegroups.com
Cc: Rik Henson <Rik.H...@mrc-cbu.cam.ac.uk>; Vladimir Litvak <litvak....@gmail.com>; Mainak Jas <main...@gmail.com>; Guiomar Niso <guioma...@ctb.upm.es>; Elizabeth Anne Bock, Ms <elizabe...@mcgill.ca>; Francois Tadel, Mr <francoi...@mcgill.ca>; Martin Pierre Cousineau, M <martin.p....@mcgill.ca>
Subject: Re: [bids-discussion] Corregistration between MEG and MRI

hi sylvain,

cheers,

satra

Sylvain.

-

EMAIL: rik.h...@mrc-cbu.cam.ac.uk
URL: http://www.mrc-cbu.cam.ac.uk/people/rik.henson/personal

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/3BC789E6-F43D-418B-96B3-D9BBA7AF2C69%40mcgill.ca.
For more options, visit https://groups.google.com/d/optout.

--
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.

Vladimir Litvak

unread,

Feb 21, 2018, 5:47:27 AM2/21/18

to Rik Henson, Satrajit Ghosh, bids-di...@googlegroups.com, Mainak Jas, Guiomar Niso, Elizabeth Anne Bock, Ms, Francois Tadel, Mr, Martin Pierre Cousineau, M

Hi Rik,

I think Satra meant that some people who are very meticulous about
their coregistration would go as far as identifying the landmarks
every time anew (e.g. based on photos). This could be supported in my
original proposal (the one with 3 steps) but at this stage perhaps we
should not keep going in circles. One possible workaround for this
case is to include in the LandmarkCoordinates field several sets of
landmarks with slightly different names.

Best,

Vladimir

On Wed, Feb 21, 2018 at 7:46 AM, Rik Henson

Rik Henson

unread,

Feb 21, 2018, 8:29:36 AM2/21/18

to Vladimir Litvak, Satrajit Ghosh, bids-di...@googlegroups.com, Mainak Jas, Guiomar Niso, Elizabeth Anne Bock, Ms, Francois Tadel, Mr, Martin Pierre Cousineau, M

Perhaps this then relates to Chris's question about whether they should be called AnatomicalLandmarks or Landmarks. I would argue that, if they are true *anatomical* landmarks, then they should NOT vary with how they happened to be digitised in a particular MEG session - ie they would be identified once on an MRI by an expert, and would not change ever again.

If however the particular choice of landmark varied from MEG session to MEG session (eg in one MEG session, a researcher digitised some landmarks based on a photo (with no standard anatomical label), and on a different MEG session, a researcher digitised some *different* landmarks), then I suppose it might be more efficient to put in each MEG JSON separately (in MRI coordinates, with labels matching those for the same points in MEG/digitised coordinates in same JSON, and ideally linked to a photo or something that defines them non-anatomically). But I would think such "landmarks" become more like headpoints - ie depend on MEG session, but are not anatomically defined, so should not appear in the MRI JSON. (Besides, is this done often? I thought most labs used anatomical landmarks, if any?).

For this reason, perhaps we should keep "AnatomicalLandmarks" in the MRI JSON, to clarify that they are *absolute anatomical locations* that do not depend on which particular landmarks were digitised in a specific MEG session?

R

PS Note that different *anatomical* landmarks could be used in different MEG sessions, ie the MRI JSON could contain a superset of anatomical locations, of which at least 3 used for any particular MEG session.

Guiomar Niso

unread,

Feb 21, 2018, 10:08:44 AM2/21/18

to Rik Henson, Vladimir Litvak, Satrajit Ghosh, bids-di...@googlegroups.com, Mainak Jas, Elizabeth Anne Bock, Ms, Francois Tadel, Mr, Martin Pierre Cousineau, M

Hi!

Sorry I had missed this chain of mails.

I think we almost agree on this now:

As Rik says, there's a set of true *anatomical* landmarks, that should go next to the MRI JSON.

(And in case a more meticulous analysis is required, we can always add new ones to the set, as Vladimir says).

Another thing I'm getting some doubts now, is that we have removed the anatomical landmark fields in the MEG JSON. Shouldn't we need to make them back, to have these points in the MEG Coordinate system and be able to make the correspondence with the MRI fields? or should all (both coils and landmarks) be stored there in HeadCoilds fields??

Cheers,

Julia Guiomar Niso Galán, PhD

Montreal Neurological Institute, McGill University

McConnell Brain Imaging Centre

guioma...@mcgill.ca

> <bids-discussion@googlegroups.com>; Chris Gorgolewski
> (krzysztof.gorgolewski@gmail.com) <krzysztof.gorgolewski@gmail.com>;

> <bids-discussion@googlegroups.com>; Chris Gorgolewski
> (krzysztof.gorgolewski@gmail.com) <krzysztof.gorgolewski@gmail.com>

> 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/3BC789E6-F43D-418B-96B3-D9BBA7AF2C69%40mcgill.ca.
> For more options, visit https://groups.google.com/d/optout.
>
>
>
> --
> 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.

Rik Henson

unread,

Feb 21, 2018, 10:12:46 AM2/21/18

to Guiomar Niso, Vladimir Litvak, Satrajit Ghosh, bids-di...@googlegroups.com, Mainak Jas, Elizabeth Anne Bock, Ms, Francois Tadel, Mr, Martin Pierre Cousineau, M

Ø Another thing I'm getting some doubts now, is that we have removed the anatomical landmark fields in the MEG JSON. Shouldn't we need to make them back, to have these points in the MEG Coordinate system and be able to make the correspondence with the MRI fields? or should all (both coils and landmarks) be stored there in HeadCoilds fields??

I’m sorry if I missed this, but I have been assuming that the same set of anatomical points (eg NAS, LPA, RPA) appear in both the MRI JSON and the MEG JSON. In the MRI JSON they are in MRI space (manually identified by an expert); in the MEG JSON, they are in the “MEG” space (ie same space as headpoints, coils, etc) (digitised during MEG experiment). Is that not currently the case?

Julia Guiomar Niso Galán, PhD

Montreal Neurological Institute, McGill University

McConnell Brain Imaging Centre

guioma...@mcgill.ca

2018-02-21 14:29 GMT+01:00 Rik Henson <Rik.H...@mrc-cbu.cam.ac.uk>:

Perhaps this then relates to Chris's question about whether they should be called AnatomicalLandmarks or Landmarks. I would argue that, if they are true *anatomical* landmarks, then they should NOT vary with how they happened to be digitised in a particular MEG session - ie they would be identified once on an MRI by an expert, and would not change ever again.

If however the particular choice of landmark varied from MEG session to MEG session (eg in one MEG session, a researcher digitised some landmarks based on a photo (with no standard anatomical label), and on a different MEG session, a researcher digitised some *different* landmarks), then I suppose it might be more efficient to put in each MEG JSON separately (in MRI coordinates, with labels matching those for the same points in MEG/digitised coordinates in same JSON, and ideally linked to a photo or something that defines them non-anatomically). But I would think such "landmarks" become more like headpoints - ie depend on MEG session, but are not anatomically defined, so should not appear in the MRI JSON. (Besides, is this done often? I thought most labs used anatomical landmarks, if any?).

For this reason, perhaps we should keep "AnatomicalLandmarks" in the MRI JSON, to clarify that they are *absolute anatomical locations* that do not depend on which particular landmarks were digitised in a specific MEG session?

R

PS Note that different *anatomical* landmarks could be used in different MEG sessions, ie the MRI JSON could contain a superset of anatomical locations, of which at least 3 used for any particular MEG session.

-----Original Message-----
From: Vladimir Litvak [mailto:litvak....@gmail.com]
Sent: 21 February 2018 10:47
To: Rik Henson <Rik.H...@mrc-cbu.cam.ac.uk>

Cc: Satrajit Ghosh <sa...@mit.edu>; bids-di...@googlegroups.com; Mainak Jas <main...@gmail.com>; Guiomar Niso <guioma...@ctb.upm.es>; Elizabeth Anne Bock, Ms <elizabe...@mcgill.ca>; Francois Tadel, Mr <francoi...@mcgill.ca>; Martin Pierre Cousineau, M <martin.p....@mcgill.ca>

> <bids-di...@googlegroups.com>; Chris Gorgolewski
> (krzysztof....@gmail.com) <krzysztof....@gmail.com>;

> <bids-di...@googlegroups.com>; Chris Gorgolewski
> (krzysztof....@gmail.com) <krzysztof....@gmail.com>

> 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/3BC789E6-F43D-418B-96B3-D9BBA7AF2C69%40mcgill.ca.
> For more options, visit https://groups.google.com/d/optout.
>
>
>
> --
> 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.

Guiomar Niso

unread,

Feb 21, 2018, 10:18:39 AM2/21/18

to Rik Henson, Vladimir Litvak, Satrajit Ghosh, bids-di...@googlegroups.com, Mainak Jas, Elizabeth Anne Bock, Ms, Francois Tadel, Mr, Martin Pierre Cousineau, M

That should be like this, right? but I think we have accidentally removed them, unless they are supposed to go together with the head coils.

I'm going to add them again.

Julia Guiomar Niso Galán, PhD

Montreal Neurological Institute, McGill University

McConnell Brain Imaging Centre

guioma...@mcgill.ca

2018-02-21 16:12 GMT+01:00 Rik Henson <Rik.H...@mrc-cbu.cam.ac.uk>:

Ø Another thing I'm getting some doubts now, is that we have removed the anatomical landmark fields in the MEG JSON. Shouldn't we need to make them back, to have these points in the MEG Coordinate system and be able to make the correspondence with the MRI fields? or should all (both coils and landmarks) be stored there in HeadCoilds fields??

I’m sorry if I missed this, but I have been assuming that the same set of anatomical points (eg NAS, LPA, RPA) appear in both the MRI JSON and the MEG JSON. In the MRI JSON they are in MRI space (manually identified by an expert); in the MEG JSON, they are in the “MEG” space (ie same space as headpoints, coils, etc) (digitised during MEG experiment). Is that not currently the case?

Julia Guiomar Niso Galán, PhD

Montreal Neurological Institute, McGill University

McConnell Brain Imaging Centre

guioma...@mcgill.ca

2018-02-21 14:29 GMT+01:00 Rik Henson <Rik.H...@mrc-cbu.cam.ac.uk>:

Perhaps this then relates to Chris's question about whether they should be called AnatomicalLandmarks or Landmarks. I would argue that, if they are true *anatomical* landmarks, then they should NOT vary with how they happened to be digitised in a particular MEG session - ie they would be identified once on an MRI by an expert, and would not change ever again.

If however the particular choice of landmark varied from MEG session to MEG session (eg in one MEG session, a researcher digitised some landmarks based on a photo (with no standard anatomical label), and on a different MEG session, a researcher digitised some *different* landmarks), then I suppose it might be more efficient to put in each MEG JSON separately (in MRI coordinates, with labels matching those for the same points in MEG/digitised coordinates in same JSON, and ideally linked to a photo or something that defines them non-anatomically). But I would think such "landmarks" become more like headpoints - ie depend on MEG session, but are not anatomically defined, so should not appear in the MRI JSON. (Besides, is this done often? I thought most labs used anatomical landmarks, if any?).

For this reason, perhaps we should keep "AnatomicalLandmarks" in the MRI JSON, to clarify that they are *absolute anatomical locations* that do not depend on which particular landmarks were digitised in a specific MEG session?

R

PS Note that different *anatomical* landmarks could be used in different MEG sessions, ie the MRI JSON could contain a superset of anatomical locations, of which at least 3 used for any particular MEG session.

-----Original Message-----
From: Vladimir Litvak [mailto:litvak.vladimir@gmail.com]
Sent: 21 February 2018 10:47
To: Rik Henson <Rik.H...@mrc-cbu.cam.ac.uk>

Cc: Satrajit Ghosh <sa...@mit.edu>; bids-discussion@googlegroups.com; Mainak Jas <main...@gmail.com>; Guiomar Niso <guioma...@ctb.upm.es>; Elizabeth Anne Bock, Ms <elizabe...@mcgill.ca>; Francois Tadel, Mr <francoi...@mcgill.ca>; Martin Pierre Cousineau, M <martin.p....@mcgill.ca>

> <bids-discussion@googlegroups.com>; Chris Gorgolewski
> (krzysztof.gorgolewski@gmail.com) <krzysztof.gorgolewski@gmail.com>;

> <bids-discussion@googlegroups.com>; Chris Gorgolewski
> (krzysztof.gorgolewski@gmail.com) <krzysztof.gorgolewski@gmail.com>

> 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/3BC789E6-F43D-418B-96B3-D9BBA7AF2C69%40mcgill.ca.
> For more options, visit https://groups.google.com/d/optout.
>
>
>
> --
> 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.

Guiomar Niso

unread,

Feb 21, 2018, 10:21:02 AM2/21/18

to Rik Henson, Vladimir Litvak, Satrajit Ghosh, bids-di...@googlegroups.com, Mainak Jas, Elizabeth Anne Bock, Ms, Francois Tadel, Mr, Martin Pierre Cousineau, M

Someone is faster than me! ;)

Julia Guiomar Niso Galán, PhD

Montreal Neurological Institute, McGill University

McConnell Brain Imaging Centre

guioma...@mcgill.ca

Sylvain Baillet, Dr

unread,

Feb 21, 2018, 10:32:20 AM2/21/18

to bids-di...@googlegroups.com, Rik Henson, Vladimir Litvak, Satrajit Ghosh, Mainak Jas, Elizabeth Anne Bock, Ms, Francois Tadel, Mr, Martin Pierre Cousineau, M

Thanks again for your suggestions folks, Satra, Rik, Vlad Guio et al.

I have tried to accommodate everyone’s righteous requests. Please see updated document.

a AnatomicalLandmarkCoordinates field is now back into the _coordsystem.json. I have also elaborated on the rationale and solutions for possible multiple instances of T1w AnatomicalLandmarkCoordinates:

In principle, these locations are those of absolute anatomical markers. However, the marking of NAS, LPA and RPA is more ambiguous than that of e.g., AC and PC. This may result in some variability in their 3-D digitization from session to session, even for the same participant. The solution to the investigator is either to duplicate the structural T1w MRI files - one per session - and specify different AnatomicalLandmarkCoordinates in voxel units in each T1w JSON. Another solution would be to use only one T1w file (RECOMMENDED) and populate the AnatomicalLandmarkCoordinates field with session-specific labels e.g., "NAS-session1": [127,213,139],"NAS-session2": [123,220,142], etc.

Thanks again!

Sylvain.

-

Sylvain Baillet, PhD

Professor, Montreal Neurological Institute

McGill University

> more

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/CAKcdbz%3DkaSekYZ9PzNnaFP96ymdcCu7CVDj41XZkUsCddVQ1Gw%40mail.gmail.com.

Lorenzo Magazzini

unread,

Feb 22, 2018, 5:51:51 AM2/22/18

to bids-di...@googlegroups.com, Rik Henson, Vladimir Litvak, Satrajit Ghosh, Mainak Jas, Elizabeth Anne Bock, Ms, Francois Tadel, Mr, Martin Pierre Cousineau, M

Hi all,

I've been following this discussion as I'm in the process of finalising a BIDS-formatted MEG dataset. Can I please ask for a clarification?

My understanding now is that AnatomicalLandmarkCoordinates should only be used to indicate clear anatomical points (NAS, LPA, RPA). This is irrespective of the actual landmarks where the fiducial coils were placed. So where are the coil locations documented?

In our lab, many researchers have systematically been placing the coils 1cm away from the actual anatomical points. In data I've seen from other labs (Yokogawa system), the centre of the forehead (marked with a capsule on the MRI) was used in place of a "nasion" fiducial. Where should I store the MRI voxel coordinates of these fiducial locations?

Sorry if I've missed this from previous emails and thanks everyone for the effort in solving this issue!

Best wishes,

Lorenzo

Lorenzo Magazzini, PhD

Research Associate

CUBRIC

Cardiff University

Cardiff CF24 4HQ

Tel: +44 (0)2920 870090

http://psych.cf.ac.uk/contactsandpeople/magazzinil.php

Cc: Satrajit Ghosh <sa...@mit.edu>; bids-discu...@googlegroups.com; Mainak Jas <main...@gmail.com>; Guiomar Niso <guioma...@ctb.upm.es>; Elizabeth Anne Bock, Ms <elizabe...@mcgill.ca>; Francois Tadel, Mr <francoi...@mcgill.ca>; Martin Pierre Cousineau, M <martin.p....@mcgill.ca>

Subject: Re: [bids-discussion] Corregistration between MEG and MRI

Hi Rik,

I think Satra meant that some people who are very meticulous about their coregistration would go as far as identifying the landmarks every time anew (e.g. based on photos). This could be supported in my original proposal (the one with 3 steps) but at this stage perhaps we should not keep going in circles. One possible workaround for this case is to include in the LandmarkCoordinates field several sets of landmarks with slightly different names.

Best,

Vladimir

On Wed, Feb 21, 2018 at 7:46 AM, Rik Henson <Rik.H...@mrc-cbu.cam.ac.uk> wrote:
>
>
> Sorry – I don’t understand why you would need new anatomical landmarks
> for each MEG session in the MRI JSON (apologies if I misunderstood
> Satra’s suggestion below). These landmarks are a property of the MRI
> image, not of the MEG data. They are *before* any coregistration of
> the MEG and MRI. (Note that their coordinates are likely to differ
> from the same landmarks that are digitized in the MEG JSON, for which
> there could be one JSON per MEG session to allow for different
> digitizations in different sessions). So I cannot see why there would need to be more than one set in the MRI JSON?
>
>
>
> R
>
>
>
>
>
> ---------------------------------------------------------------
>
> Professor Richard Henson
>
> Deputy Director
> MRC Cognition and Brain Sciences Unit
> University of Cambridge
>
> 15 Chaucer Road
> Cambridge, CB2 7EF
> England
>

> EMAIL: rik.henson@mrc-cbu.cam.ac.uk

> an email to bids-discussion+unsub...@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/3BC789E6-F43D-418B-96B3-D9BBA7AF2C69%40mcgill.ca.
> For more options, visit https://groups.google.com/d/optout.
>
>
>
> --
> 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+unsub...@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/4B972926-3FB1-49FF-A98F-8BBCC119063F%40mcgill.ca.
>
>
> For more options, visit https://groups.google.com/d/optout.
>
>

--
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+unsub...@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/CAKcdbz%3DkaSekYZ9PzNnaFP96ymdcCu7CVDj41XZkUsCddVQ1Gw%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

--
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/596F183A-6669-4D0F-9299-CAA556E72326%40mcgill.ca.

Rik Henson

unread,

Feb 22, 2018, 6:34:09 AM2/22/18

to Lorenzo Magazzini, bids-di...@googlegroups.com, Vladimir Litvak, Satrajit Ghosh, Mainak Jas, Elizabeth Anne Bock, Ms, Francois Tadel, Mr, Martin Pierre Cousineau, M

My suggestion would be that the MRI coordinates of fiducials marked by capsules that are visible in the MRI should be stored in the MRI JSON file, since they are “anatomical” in the sense that they are property of the MRI image itself (and can be given a label like “capsule1” that would then need to appear in the MEG JSON with its corresponding digitized MEG coordinates).

As for coils put a fixed distance from an anatomical landmark, then I would suggest that their MRI coordinates are also put in the MRI JSON, since they are still defined relative to anatomy, even if not coincident with an exact anatomical feature. It’s probably better not to give them a conventional anatomical label like “LPA”, but rather a bespoke label like “LPA1cm” – all that matters (I think) is that the same label is used in the MEG JSON file for their MEG coordinates.

R

From: Lorenzo Magazzini [mailto:magaz...@gmail.com]
Sent: 22 February 2018 10:51
To: bids-di...@googlegroups.com
Cc: Rik Henson <Rik.H...@mrc-cbu.cam.ac.uk>; Vladimir Litvak <litvak....@gmail.com>; Satrajit Ghosh <sa...@mit.edu>; Mainak Jas <main...@gmail.com>; Elizabeth Anne Bock, Ms <elizabe...@mcgill.ca>; Francois Tadel, Mr <francoi...@mcgill.ca>; Martin Pierre Cousineau, M <martin.p....@mcgill.ca>
Subject: Re: [bids-discussion] Corregistration between MEG and MRI

Hi all,

Cc: Satrajit Ghosh <sa...@mit.edu>; bids-di...@googlegroups.com; Mainak Jas <main...@gmail.com>; Guiomar Niso <guioma...@ctb.upm.es>; Elizabeth Anne Bock, Ms <elizabe...@mcgill.ca>; Francois Tadel, Mr <francoi...@mcgill.ca>; Martin Pierre Cousineau, M <martin.p....@mcgill.ca>

Subject: Re: [bids-discussion] Corregistration between MEG and MRI

Hi Rik,

I think Satra meant that some people who are very meticulous about their coregistration would go as far as identifying the landmarks every time anew (e.g. based on photos). This could be supported in my original proposal (the one with 3 steps) but at this stage perhaps we should not keep going in circles. One possible workaround for this case is to include in the LandmarkCoordinates field several sets of landmarks with slightly different names.

Best,

Vladimir

On Wed, Feb 21, 2018 at 7:46 AM, Rik Henson <Rik.H...@mrc-cbu.cam.ac.uk> wrote:
>
>
> Sorry – I don’t understand why you would need new anatomical landmarks
> for each MEG session in the MRI JSON (apologies if I misunderstood
> Satra’s suggestion below). These landmarks are a property of the MRI
> image, not of the MEG data. They are *before* any coregistration of
> the MEG and MRI. (Note that their coordinates are likely to differ
> from the same landmarks that are digitized in the MEG JSON, for which
> there could be one JSON per MEG session to allow for different
> digitizations in different sessions). So I cannot see why there would need to be more than one set in the MRI JSON?
>
>
>
> R
>
>
>
>
>
> ---------------------------------------------------------------
>
> Professor Richard Henson
>
> Deputy Director
> MRC Cognition and Brain Sciences Unit
> University of Cambridge
>
> 15 Chaucer Road
> Cambridge, CB2 7EF
> England
>

> EMAIL: rik.h...@mrc-cbu.cam.ac.uk

> URL: http://www.mrc-cbu.cam.ac.uk/people/rik.henson/personal
>
> TEL +44 (0)1223 355 294 x501
> FAX +44 (0)1223 359 062
> MOB +44 (0)794 1377 345
>
> ---------------------------------------------------------------
>
>
>
>
>
> From: Sylvain Baillet, Dr [mailto:sylvain...@mcgill.ca]
> Sent: 20 February 2018 14:19
> To: Rik Henson <Rik.H...@mrc-cbu.cam.ac.uk>
> Cc: Vladimir Litvak <litvak....@GMAIL.COM>; bids-discussion

> <bids-di...@googlegroups.com>; Chris Gorgolewski
> (krzysztof....@gmail.com) <krzysztof....@gmail.com>;

> EMAIL: rik.h...@mrc-cbu.cam.ac.uk

> URL: http://www.mrc-cbu.cam.ac.uk/people/rik.henson/personal
>
> TEL +44 (0)1223 355 294 x501
> FAX +44 (0)1223 359 062
> MOB +44 (0)794 1377 345
>
> ---------------------------------------------------------------
>
>
>
>
>
> From: Sylvain Baillet, Dr [mailto:sylvain...@mcgill.ca]
> Sent: 19 February 2018 15:48
> To: Vladimir Litvak <litvak....@GMAIL.COM>
> Cc: Rik Henson <Rik.H...@mrc-cbu.cam.ac.uk>; bids-discussion

> <bids-di...@googlegroups.com>; Chris Gorgolewski
> (krzysztof....@gmail.com) <krzysztof....@gmail.com>

> EMAIL: rik.h...@mrc-cbu.cam.ac.uk

> 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/3BC789E6-F43D-418B-96B3-D9BBA7AF2C69%40mcgill.ca.
> For more options, visit https://groups.google.com/d/optout.
>
>
>
> --
> 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/4B972926-3FB1-49FF-A98F-8BBCC119063F%40mcgill.ca.
>
>
> For more options, visit https://groups.google.com/d/optout.
>
>

--
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/CAKcdbz%3DkaSekYZ9PzNnaFP96ymdcCu7CVDj41XZkUsCddVQ1Gw%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

--
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.

Lorenzo Magazzini

unread,

Feb 22, 2018, 6:44:58 AM2/22/18

to Rik Henson, bids-di...@googlegroups.com, Vladimir Litvak, Satrajit Ghosh, Mainak Jas, Elizabeth Anne Bock, Ms, Francois Tadel, Mr, Martin Pierre Cousineau, M

Thanks, Rik. I guess my confusion stems from the fact that the three fiducial coils are often referred to as 'nas', 'lpa', 'rpa' in MEG recordings as well as in analysis software, even when they're not placed on the actual nasion and pre-auricular anatomical landmarks.

I think you're right in saying that consistency between MEG and MRI labels should avoid any confusion.

Best,

Lorenzo

Lorenzo Magazzini, PhD

Research Associate

CUBRIC

Cardiff University

Cardiff CF24 4HQ

Tel: +44 (0)2920 870090

http://psych.cf.ac.uk/contactsandpeople/magazzinil.php

Cc: Satrajit Ghosh <sa...@mit.edu>; bids-discu...@googlegroups.com; Mainak Jas <main...@gmail.com>; Guiomar Niso <guioma...@ctb.upm.es>; Elizabeth Anne Bock, Ms <elizabe...@mcgill.ca>; Francois Tadel, Mr <francoi...@mcgill.ca>; Martin Pierre Cousineau, M <martin.p....@mcgill.ca>

Subject: Re: [bids-discussion] Corregistration between MEG and MRI

Hi Rik,

I think Satra meant that some people who are very meticulous about their coregistration would go as far as identifying the landmarks every time anew (e.g. based on photos). This could be supported in my original proposal (the one with 3 steps) but at this stage perhaps we should not keep going in circles. One possible workaround for this case is to include in the LandmarkCoordinates field several sets of landmarks with slightly different names.

Best,

Vladimir

On Wed, Feb 21, 2018 at 7:46 AM, Rik Henson <Rik.H...@mrc-cbu.cam.ac.uk> wrote:
>
>
> Sorry – I don’t understand why you would need new anatomical landmarks
> for each MEG session in the MRI JSON (apologies if I misunderstood
> Satra’s suggestion below). These landmarks are a property of the MRI
> image, not of the MEG data. They are *before* any coregistration of
> the MEG and MRI. (Note that their coordinates are likely to differ
> from the same landmarks that are digitized in the MEG JSON, for which
> there could be one JSON per MEG session to allow for different
> digitizations in different sessions). So I cannot see why there would need to be more than one set in the MRI JSON?
>
>
>
> R
>
>
>
>
>
> ---------------------------------------------------------------
>
> Professor Richard Henson
>
> Deputy Director
> MRC Cognition and Brain Sciences Unit
> University of Cambridge
>
> 15 Chaucer Road
> Cambridge, CB2 7EF
> England
>

> EMAIL: rik.henson@mrc-cbu.cam.ac.uk

> URL: http://www.mrc-cbu.cam.ac.uk/people/rik.henson/personal
>
> TEL +44 (0)1223 355 294 x501
> FAX +44 (0)1223 359 062
> MOB +44 (0)794 1377 345
>
> ---------------------------------------------------------------
>
>
>
>
>
> From: Sylvain Baillet, Dr [mailto:sylvain.baillet@mcgill.ca]
> Sent: 20 February 2018 14:19
> To: Rik Henson <Rik.H...@mrc-cbu.cam.ac.uk>
> Cc: Vladimir Litvak <litvak....@GMAIL.COM>; bids-discussion

> <bids-discussion@googlegroups.com>; Chris Gorgolewski
> (krzysztof.gorgolewski@gmail.com) <krzysztof.gorgolewski@gmail.com>;

> EMAIL: rik.henson@mrc-cbu.cam.ac.uk

> URL: http://www.mrc-cbu.cam.ac.uk/people/rik.henson/personal
>
> TEL +44 (0)1223 355 294 x501
> FAX +44 (0)1223 359 062
> MOB +44 (0)794 1377 345
>
> ---------------------------------------------------------------
>
>
>
>
>
> From: Sylvain Baillet, Dr [mailto:sylvain.baillet@mcgill.ca]
> Sent: 19 February 2018 15:48
> To: Vladimir Litvak <litvak....@GMAIL.COM>
> Cc: Rik Henson <Rik.H...@mrc-cbu.cam.ac.uk>; bids-discussion

> <bids-discussion@googlegroups.com>; Chris Gorgolewski
> (krzysztof.gorgolewski@gmail.com) <krzysztof.gorgolewski@gmail.com>

> EMAIL: rik.henson@mrc-cbu.cam.ac.uk

> an email to bids-discussion+unsub...@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/3BC789E6-F43D-418B-96B3-D9BBA7AF2C69%40mcgill.ca.
> For more options, visit https://groups.google.com/d/optout.
>
>
>
> --
> 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+unsub...@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/4B972926-3FB1-49FF-A98F-8BBCC119063F%40mcgill.ca.
>
>
> For more options, visit https://groups.google.com/d/optout.
>
>

--
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+unsub...@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/CAKcdbz%3DkaSekYZ9PzNnaFP96ymdcCu7CVDj41XZkUsCddVQ1Gw%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

--
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.

Vladimir Litvak

unread,

Feb 22, 2018, 6:57:43 AM2/22/18

to Lorenzo Magazzini, Rik Henson, bids-di...@googlegroups.com, Satrajit Ghosh, Mainak Jas, Elizabeth Anne Bock, Ms, Francois Tadel, Mr, Martin Pierre Cousineau, M

Hi all,

One possibility that was part of my 3-step solution is that the
fiducials are specified in MEG json's with the corresponding labels
from MRI json rather than numbers. In your example it could be e.g.:

{
nas, NAS;
lpa, LPA1cm
rpa, RPA1cm
}

If that's the only thing we allow for MEG json it would solve the
problem of raw vs. derived because this information does not change.
It would also allow linking one MEG recording to several different
structurals as long as the landmark names are the same in all of them.
Finally if one changes something about the MRI, only its own JSON will
have to be updated and there is no need to keep track of what MEG
datasets are linked to it. How does that sound?

Vladimir

>> Cc: Satrajit Ghosh <sa...@mit.edu>; bids-di...@googlegroups.com;

>> > EMAIL: rik.h...@mrc-cbu.cam.ac.uk

>> > URL: http://www.mrc-cbu.cam.ac.uk/people/rik.henson/personal
>> >
>> > TEL +44 (0)1223 355 294 x501
>> > FAX +44 (0)1223 359 062
>> > MOB +44 (0)794 1377 345
>> >
>> > ---------------------------------------------------------------
>> >
>> >
>> >
>> >
>> >
>> > From: Sylvain Baillet, Dr [mailto:sylvain...@mcgill.ca]
>> > Sent: 20 February 2018 14:19
>> > To: Rik Henson <Rik.H...@mrc-cbu.cam.ac.uk>
>> > Cc: Vladimir Litvak <litvak....@GMAIL.COM>; bids-discussion

>> > <bids-di...@googlegroups.com>; Chris Gorgolewski
>> > (krzysztof....@gmail.com) <krzysztof....@gmail.com>;

>> > EMAIL: rik.h...@mrc-cbu.cam.ac.uk

>> > URL: http://www.mrc-cbu.cam.ac.uk/people/rik.henson/personal
>> >
>> > TEL +44 (0)1223 355 294 x501
>> > FAX +44 (0)1223 359 062
>> > MOB +44 (0)794 1377 345
>> >
>> > ---------------------------------------------------------------
>> >
>> >
>> >
>> >
>> >
>> > From: Sylvain Baillet, Dr [mailto:sylvain...@mcgill.ca]
>> > Sent: 19 February 2018 15:48
>> > To: Vladimir Litvak <litvak....@GMAIL.COM>
>> > Cc: Rik Henson <Rik.H...@mrc-cbu.cam.ac.uk>; bids-discussion

>> > <bids-di...@googlegroups.com>; Chris Gorgolewski
>> > (krzysztof....@gmail.com) <krzysztof....@gmail.com>

>> > EMAIL: rik.h...@mrc-cbu.cam.ac.uk

>> > 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/3BC789E6-F43D-418B-96B3-D9BBA7AF2C69%40mcgill.ca.
>> > For more options, visit https://groups.google.com/d/optout.
>> >
>> >
>> >
>> > --
>> > 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/4B972926-3FB1-49FF-A98F-8BBCC119063F%40mcgill.ca.
>> >
>> >
>> > For more options, visit https://groups.google.com/d/optout.
>> >
>> >
>>
>>
>>
>>
>>
>>
>>
>>
>>
>> --
>> 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/CAKcdbz%3DkaSekYZ9PzNnaFP96ymdcCu7CVDj41XZkUsCddVQ1Gw%40mail.gmail.com.
>> For more options, visit https://groups.google.com/d/optout.
>>
>>
>>
>> --
>> 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.

Sylvain Baillet, Dr

unread,

Feb 22, 2018, 8:08:50 AM2/22/18

to bids-di...@googlegroups.com, Lorenzo Magazzini, Rik Henson, Satrajit Ghosh, Mainak Jas, Elizabeth Anne Bock, Ms, Francois Tadel, Mr, Martin Pierre Cousineau, M

Hi all:

Lorenzo: the head coil coordinates are specified in the MEG _coordsystem.json (see: Fiducials (HLC/HPI Coils) section).

I understand that for all manufacturers, the MEG coordinate system is always relative to the head and defined wrt to the actual anatomical landmarks, regardless of the location of the coils (eg, CTF imposes the coils ot be placed at the anatomical landmark locations, not Elekta). Then only the anatomical landmarks need to be marked in the MR, not the coils.

Best,

Sylvain.

-

Sylvain Baillet, PhD

Professor, Montreal Neurological Institute

McGill University

> more

To view this discussion on the web visit https://groups.google.com/d/msgid/bids-discussion/CACq1BhLYWcRLq-MOkWmc5V4QocxqJskwEDjTY0b0Xqr0YBbtiw%40mail.gmail.com.

Lorenzo Magazzini

unread,

Feb 22, 2018, 12:14:54 PM2/22/18

to Sylvain Baillet, Dr, bids-di...@googlegroups.com, Rik Henson, Satrajit Ghosh, Mainak Jas, Elizabeth Anne Bock, Ms, Francois Tadel, Mr, Martin Pierre Cousineau, M

Hi Sylvain,

Thanks - I really hope that the following will not just add confusion to the discussion...

Lorenzo: the head coil coordinates are specified in the MEG _coordsystem.json (see: Fiducials (HLC/HPI Coils) section).

Let me check if I understand correctly:

HeadCoilCoordinates, in the *_coordsystem.json file accompanying the MEG data, refers to the coordinates of the fiducial coils, expressed in the coordinate system specified in HeadCoilCoordinateSystem (for example, CTFGradiometer, NeuromagElekta, etc.).

AnatomicalLandmarkCoordinates, in the same *_coordsystem.json file, refers to the coordinates of the actual anatomical landmarks (typically NAS, LPA, RPA), expressed in the coordinate system specified in AnatomicalLandmarkCoordinateSystem (and I guess ideally in the same coordinate system as HeadCoilCoordinateSystem).
Crucially, these coordinates might be unavailable to the researcher (e.g., if the head coils were not placed on the actual anatomical landmarks and the anatomical landmarks were not 3-D digitized).

What is still not entirely clear to me is whether these two sets of coordinates (HeadCoilCoordinates and AnatomicalLandmarkCoordinates) are stored also in the json file accompanying the anatomical MRI file.
If this is the case, then it's crucial that the keys are labelled consistently with the MEG json (as Rik said).
If this is not the case, then please let me know where the HeadCoilCoordinates in voxel units should be stored.

I understand that for all manufacturers, the MEG coordinate system is always relative to the head and defined wrt to the actual anatomical landmarks, regardless of the location of the coils (eg, CTF imposes the coils ot be placed at the anatomical landmark locations, not Elekta).
Then only the anatomical landmarks need to be marked in the MR, not the coils.

I thought this was not the case. For CTF, for example, I thought the coordinates system was defined based on the three fiducial coils, irrespective of the actual anatomical landmarks. The coils represent the NAS, LPA and RPA landmarks, but in practice are often placed close to these points (e.g., NAS between the eyebrows), for convenience.

If only HeadCoilCoordinates (and not AnatomicalLandmarkCoordinates) are available to the researcher, then it's crucial that the corresponding voxel coordinates are marked on the associated anatomical MRI.

Finally, I just wanted to mention that in raw MEG data from the Yokogawa system, the location of the MEG sensors is not expressed relative to the fiducial coils (i.e. the head). Or at least my understanding is that MEG sensors and fiducial coils coordinates need to be realigned to a head coordinate system by software processing.

Best wishes,

Lorenzo

Cc: Satrajit Ghosh <sa...@mit.edu>; bids-discussion@googlegroups.com;

<bids-discussion@googlegroups.com>; Chris Gorgolewski
(krzysztof.gorgolewski@gmail.com) <krzysztof.gorgolewski@gmail.com>;

<bids-discussion@googlegroups.com>; Chris Gorgolewski
(krzysztof.gorgolewski@gmail.com) <krzysztof.gorgolewski@gmail.com>

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/3BC789E6-F43D-418B-96B3-D9BBA7AF2C69%40mcgill.ca.
For more options, visit https://groups.google.com/d/optout.

--
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/4B972926-3FB1-49FF-A98F-8BBCC119063F%40mcgill.ca.

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

--
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/CAKcdbz%3DkaSekYZ9PzNnaFP96ymdcCu7CVDj41XZkUsCddVQ1Gw%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

--
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/596F183A-6669-4D0F-9299-CAA556E72326%40mcgill.ca.

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

--
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.

Guiomar Niso

unread,

Feb 22, 2018, 4:50:45 PM2/22/18

to bids-discussion, Sylvain Baillet, Dr, Rik Henson, Satrajit Ghosh, Mainak Jas, Elizabeth Anne Bock, Ms, Francois Tadel, Mr, Martin Pierre Cousineau, M

Hi Lorenzo,

Exactly:

- Anatomical Landmarks (in MEG system) --> AnatomicalLandmarkCoordinates

- Coils (in MEG system) --> HeadCoilCoordinates

Corresponding Anatomical Landmarks (in MRI system) go in the MRI JSON file: AnatomicalLandmarkCoordinates

This field could contain as many points as we want, i.e. it could be a bigger set than just the three needed in MEG.

For the correspondence between MEG and MRI coordinates, we'll have two options:
1) Ensure (with the validator) that any MEG set of anatomical landmarks is defined in the MRI one.
2) Add a field in the MEG JSON with the labels as Vladimir suggests. This has the advantage that maybe is a more methodic solution. This field will need to be checked by the validator as well.

In any case, as you say, there may be groups that don't record anatomical points or that they record different ones (not following the recommended practices).

For this specific case, as Rik was suggesting, the marked coils could be used both as coils and as anatomical landmarks, being careful to label them properly when used as anatomical landmarks ("LPA1cm" is a good example).

For the Yokogawa system, I think the same could apply, as long as it's indicated that the coordinates are expressed in this specific MEG system. Then the processing tools will need to take care of these details you mention internally when they perform the co registration.

I hope this clarifies.

Cheers!

Julia Guiomar Niso Galán, PhD

Montreal Neurological Institute, McGill University

McConnell Brain Imaging Centre

guioma...@mcgill.ca

To view this discussion on the web visit https://groups.google.com/d/msgid/bids-discussion/CABQiM1RkXcQuRW43bacqzTyvyyL69Jzj8ONVy3PPzrA1%2BP0Png%40mail.gmail.com.

Lorenzo Magazzini

unread,

Feb 23, 2018, 5:53:49 AM2/23/18

to bids-di...@googlegroups.com, Sylvain Baillet, Dr, Rik Henson, Satrajit Ghosh, Mainak Jas, Elizabeth Anne Bock, Ms, Francois Tadel, Mr, Martin Pierre Cousineau, M

HI Guiomar,

Yes! much clearer now. Thank you everyone and sorry if it took me a while to catch up with these different options.

Best wishes,

Lorenzo

Lorenzo Magazzini, PhD

Research Associate

CUBRIC

Cardiff University

Cardiff CF24 4HQ

Tel: +44 (0)2920 870090

http://psych.cf.ac.uk/contactsandpeople/magazzinil.php

To view this discussion on the web visit https://groups.google.com/d/msgid/bids-discussion/CAKcdbznLZ3TgxSQru8QYFhDuTJKRWvFvsnhwPEdfqq0Ow%3D6ONg%40mail.gmail.com.

Sylvain Baillet, Dr

unread,

Feb 23, 2018, 10:00:14 AM2/23/18

to Lorenzo Magazzini, bids-discussion, Rik Henson, Satrajit Ghosh, Mainak Jas, Elizabeth Anne Bock, Ms, Francois Tadel, Mr, Martin Pierre Cousineau, M

No pb Lorenzo - glad this is clear now.

and thanks again everyone for the input.

these multimodal coreg things are not as simple as they seem.

Chris: the MEG-BIDS specs 1.0 are finalized now.

Best,

Sylvain.

-

Chris Gorgolewski

unread,

Feb 23, 2018, 1:29:48 PM2/23/18

to bids-discussion, Lorenzo Magazzini, Rik Henson, Satrajit Ghosh, Mainak Jas, Elizabeth Anne Bock, Ms, Francois Tadel, Mr, Martin Pierre Cousineau, M

Ok thanks. Exciting times!

Just to double check - do your suggestion to the MR metadata related to landmarks in the current (main) BIDS draft need any corrections?

Next step are the examples - Guimar/Mainiak what is ready to merge? I see three different PRs open...

I will start merging the google documents. I will let you know when it is done so we could make the MEG extension draft read only and redirect people to the main document.

Best,

Chris

To view this discussion on the web visit https://groups.google.com/d/msgid/bids-discussion/00484115-6CA4-4A3C-BE4F-86E9051BF6D9%40mcgill.ca.

Mainak Jas

unread,

Feb 24, 2018, 9:23:58 PM2/24/18

to Chris Gorgolewski, bids-discussion, Lorenzo Magazzini, Rik Henson, Satrajit Ghosh, Elizabeth Anne Bock, Ms, Francois Tadel, Mr, Martin Pierre Cousineau, M, Alexandre Gramfort

Chris,

I would merge this PR if it's good to go for Guio: https://github.com/INCF/BIDS-examples/pull/51

I'll be able to get back to the validator PR next week. There are some regexps to fix, and a few updates that Guio informed me about.

I suggest freezing the MEG-BIDS draft at some point, as I still see many new suggestions/improvements everyday. The validator would work for this "version" of the draft.

Mainak

Sylvain Baillet, Dr

unread,

Feb 25, 2018, 6:44:08 PM2/25/18

to bids-di...@googlegroups.com, Chris Gorgolewski, Lorenzo Magazzini, Rik Henson, Satrajit Ghosh, Elizabeth Anne Bock, Ms, Francois Tadel, Mr, Martin Pierre Cousineau, M, Alexandre Gramfort

Thanks Mainak:

We’re done with the inaugural version of MEG-BIDS.

Please proceed with the validator as we need that part done to share the MEG-BIDS data samples and send the revised version of the manuscript.

Thank you,

Sylvain.

-

Sylvain Baillet, PhD

Professor, Montreal Neurological Institute

McGill University

> more

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/CANzrpG2arwyS_nHgx7OoxbDv_GbvmFLK12jNWOi%2BRtSBtuzrng%40mail.gmail.com.

Guiomar Niso

unread,

Feb 26, 2018, 9:25:02 PM2/26/18

to bids-discussion, Chris Filo Gorgolewski, Lorenzo Magazzini, Rik Henson, Satrajit Ghosh, Elizabeth Anne Bock, Ms, François Tadel, Martin Pierre Cousineau, M, Alexandre Gramfort

Thanks Mainak!!

Yes, go ahead, please, and merge PR #51 :)

Let's consider this is the definitive version of the specs the validator has to work for. Chris will merge it soon. If we make further changes, we assume they won't be implemented in the validator yet.

Thanks a lot!

Cheers,

El 25 feb. 2018 3:23, "Mainak Jas" <main...@gmail.com> escribió:

Chris,

I would merge this PR if it's good to go for Guio: https://github.com/INCF/BIDS-examples/pull/51

I'll be able to get back to the validator PR next week. There are some regexps to fix, and a few updates that Guio informed me about.

I suggest freezing the MEG-BIDS draft at some point, as I still see many new suggestions/improvements everyday. The validator would work for this "version" of the draft.

Mainak

To view this discussion on the web visit https://groups.google.com/d/msgid/bids-discussion/CANzrpG2arwyS_nHgx7OoxbDv_GbvmFLK12jNWOi%2BRtSBtuzrng%40mail.gmail.com.

Reply all

Reply to author

Forward