[FieldTrip] Running the ICA method to remove EOG/ECG artifact on Neuromag(*.fif) data

[Panagiotis Tsiatsis]

On 4/13/2012 8:50 AM, Alexandre Gramfort wrote:
> but be careful when applying ICA on neuromag data as
> channels have different types and you have signals
> with 2 orders of magnitude difference between gradiometers
> and magnetometers.
>
> I am curious to know if fieldtrip handles this properly? and if so how.
Hi all,

I am also working with CTF MEG data. What I ve been basically doing up 
to now was to split the datasets in two subsets: one containing only the 
MEGGRAD sensors (let s call it A)  and another containing all the rest 
(let it be B). Then I would apply ICA only on the set containing the 
main MEG sensors, subtract the EOG-related components and re-construct 
the data without them (A'). Finally, I would merge A' with B. This works 
fine and the results that I am getting are convincing, the only problem 
is that of course the reference channels are still contaminated. I don t 
know if this would be a problem in subsequent analysis, ie source 
localization, maybe somebody who has more experience further down this 
road could comment on that.


Best,
Panagiotis

-- 
Panagiotis S. Tsiatsis
Max Planck Institute for Biological Cybernetics
Cognitive NeuroImaging Group
Tuebingen, Germany