[FieldTrip] Pre-processing on orthogonal planar gradiometers?

[jan-mathijs schoffelen]

Hi Niels,

It's perfectly fine to do the ICA etc on the horizontal and vertical planar gradients together. The reason being that the topographies of the artifacts cause (just like the neural sources) gradient fields with arbitrary orientation in 3D space, for which you need both the horizontal and vertical gradient channels to optimally capture this. For that matter it is even better to use them both in the same ICA step. It's absolutely forbidden to first do combineplanar and then try to do ICA on the combined data, because after the combination step (which is non-linear) the linear mixing assumption cannot be made anymore. Combining with the magnetometers would in principle be possible (differences in magnitude are addressed by the prewhitening step in most ICA algorithms), but I don't know how common this is in the field. Perhaps other people could comment on this.

Best wishes,
Jan-Mathijs

On Nov 13, 2013, at 6:53 PM, Niels Trusbak Haumann wrote:

> Dear FieldTrip users.
> 
> I want to detect and remove EOG and ECG by using ICA. My data is from a Neuromag 306 MEG scanner with magnetometers and two orthogonal planar gradiometers.
> 
> I assume that it is recommendable to do the ICA separately on each sensor type separately, i.e. on 1. magnetomers, 2. horizontal planar gradiometers, and then 3. vertical planar gradiometers, because they measure magnetic field changes at different scales and in different directions.
> 
> Would you recommend to do ICA and other pre-processing steps separately on the horizontal and vertical planar gradiometers, or would you just combine the planar gradiometers with ft_combineplanarimmediately?
> 
> I'm worried that combining the two different gradiometer sensor types would violate underlying assumptions of the pre-processing algorithms, i.e. ICA, filters and baseline correction, because the planar gradiometer values are only positive valued estimates from the combinations of the original gradiometer measures, which do not seem to have oscillatory deflections around a baseline zero valued point.
> 
> Hope you can help.
> 
> Greetings
> Niels.
> 
> Niels Trusbak Haumann
> M.A. / PhD student
> Department of Aesthetic Studies / Center of Functionally Integrative Neuroscience
> Aarhus University / Aarhus University Hospital
> E-mail: aestnth at hum.au.dk
> 
> _______________________________________________
> fieldtrip mailing list
> fieldtrip at donders.ru.nl
> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip

Jan-Mathijs Schoffelen, MD PhD 

Donders Institute for Brain, Cognition and Behaviour, 
Centre for Cognitive Neuroimaging,
Radboud University Nijmegen, The Netherlands

Max Planck Institute for Psycholinguistics,
Nijmegen, The Netherlands

J.Schoffelen at donders.ru.nl
Telephone: +31-24-3614793

http://www.hettaligebrein.nl

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20131114/0ad47905/attachment-0002.html>