[FieldTrip] How to understand the channel location of ctf275 MEG data
Schoffelen, J.M. (Jan Mathijs)
janmathijs.schoffelen at donders.ru.nl
Tue Aug 9 08:36:25 CEST 2022
Thanks for the confirmation! For those who are still following this thread, and for reference: FieldTrip’s ft_headmovement has similar functionality as the SPM function that Vladimir mentioned.
On 5 Aug 2022, at 18:06, Vladimir Litvak <litvak.vladimir at gmail.com<mailto:litvak.vladimir at gmail.com>> wrote:
Your description of what that SPM function does is 100% correct. It's also robust to occasional poor continuous head tracking. I'm not sure if the 12cm shift is in the same example I saw but there the problem was an incorrectly recorded fiducial. That could indeed be remedied by that SPM function.
On Fri, Aug 5, 2022 at 4:27 PM Schoffelen, J.M. (Jan Mathijs) via fieldtrip <fieldtrip at science.ru.nl<mailto:fieldtrip at science.ru.nl>> wrote:
Hi Li Zhi,
I don’t understand why you think that something can (or needs to) be fixed.
You say "I am building head models for each subject hence channel location does affect the source reconstruction”. I don’t understand this. A headmodel is constructed from anatomical data (ideally), which is independent of the functional MEG data and the sensor description.
In general, a source reconstruction will be only valid if the sensor positions are expressed in the same coordinate system as the headmodel. Once this is ensured (and as a side note: for what it’s worth the origin of the coordinate system can be defined on the moon, and the distance units in lightyears, as long as its consistent), the source reconstruction will be correct. Whether it’s a good result, depends on how the subject was positioned in the MEG (e.g. far away from the sensors), but not on whether or not the subject was e.g. in sitting or supine position, or even facing backward. I don’t understand what you mean with SPM ‘fixing’ the channel locations, at least not how this will address the problem you have been sketching so far. If anything (but correct me if I’m wrong) the SPM procedure will adjust the sensor positions as they will be used for the source reconstruction to reflect the average position of the subject throughout the measurement, rather than the initial head position that was determined by the localization procedure just prior to starting data acquisition. If there’s continuous head position information in the data you are working with, you may want to check those traces for the outlier subject(s). I doubt whether this will explain away the 12 cm anterior shift (because there is not sufficient space in the MEG to move the head over such a distance).
On 5 Aug 2022, at 16:48, Schoffelen, J.M. (Jan Mathijs) <janmathijs.schoffelen at donders.ru.nl<mailto:janmathijs.schoffelen at donders.ru.nl>> wrote:
Begin forwarded message:
I am building head models for each subject hence channel location does affect the source reconstruction. Because these data were collected by another lab 10 years ago. Hence I guess I cannot know what happened to some problematic data. Now I just remove subjects with obviously wrong head models from my analysis. I know SPM can fix the channel location if continuous head location data was collected. I have tried it but the head model did not get better. May I ask if you know any other toolbox or standard pipeline for adjusting channel location for head model inversion? Many thanks.
On Fri, 5 Aug 2022 at 10:02, Schoffelen, J.M. (Jan Mathijs) <janmathijs.schoffelen at donders.ru.nl<mailto:janmathijs.schoffelen at donders.ru.nl>> wrote:
As long as you stick to a sensor level analysis, there is no need to ‘fix’ it. Of course the subject can only be included if you are sure that they were more or less normally positioned in the MEG. For this you would need to check your lab notebook (or ask the person who has collected the data).
On 4 Aug 2022, at 21:17, Zhi Li <lizhi.psych at gmail.com<mailto:lizhi.psych at gmail.com>> wrote:
Thanks for your kind suggestion. The x-coordinates of channels of the third subject are much bigger than others and I do not know why. May I ask if there is any method to fix problematic channel location after data collection? Many thanks.
On Thu, 4 Aug 2022 at 14:31, Schoffelen, J.M. (Jan Mathijs) via fieldtrip <fieldtrip at science.ru.nl<mailto:fieldtrip at science.ru.nl>> wrote:
Hi Li Zhi,
Sensor coordinates are registered to a coordinate system that is determined on a per measurement basis. This coordinate system is determined based on how the head positioning coils have been positioned on the participant’s head (typically on the nasion, and on the left/right ear). The positions of the sensors then also depend on how the participant was positioned in the MEG device, which may cause slight (but not too big) differences across participants. The first two sets of numbers that you report make sense to me, the third set of numbers (with an x-coordinate of 20 cm(?)) seems off to me. This FAQ https://www.fieldtriptoolbox.org/faq/coordsys/<https://urldefense.com/v3/__https://www.fieldtriptoolbox.org/faq/coordsys/__;!!HJOPV4FYYWzcc1jazlU!-Gw0xFiULIb9yE5AE08dtMVSFJQLNg6wMNyM-WnW_sGtfEFA5ihWeZlv__fO5h967GsT-WZm59BX8q_dCf3aUyWT_Bu6196lcb4$> explains a bit more about coordinate systems.
In general, for sensor level analyses no standardisation is done. I would double check your third subject though, because something might have gone wrong with the head localization procedure.
> On 14 Jul 2022, at 18:41, Zhi Li via fieldtrip <fieldtrip at science.ru.nl<mailto:fieldtrip at science.ru.nl>> wrote:
> I am new to MEG data and have no experience in collecting them. When I processed the ctf275 MEG data, I found that the channel location in hdr.grad.chanpos were not the same across subjects. For example, the Cartesian coordinates of MLC11 from the data of 3 subjects were [7.96, 1.67, 13.94], [8.9, 1.1, 13.68] and [20.04, -0.03, 14.33] respectively. Hence both the 3D channel location and their 2D projections are different across subjects. May I ask how these Cartesian coordinates were calculated and should we standardize them before data processing, i.e., make the channel location the same across subjects? Any suggestions would be much appreciated.
> Li Zhi
> fieldtrip mailing list
fieldtrip mailing list
fieldtrip mailing list
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the fieldtrip