[Eeglablist] Interpolation and ICA

Arnaud Delorme arno at ucsd.edu
Tue Jan 5 13:47:48 PST 2010


The ICA matrix is going to be unchanged by interpolation.
However, after interpolation, you cannot remove ICA artifact  
components from your data (otherwise only the artifactual components  
are only going to be removed only from the channels that were used to  
compute ICA). By contrast, if you remove components from your data  
prior to interpolation, this is fine.
In other words, ICA components are not interpolated,

Arno

On Jan 5, 2010, at 1:35 PM, Julie Onton wrote:

> does EEGLAB automatically fill in the EEG.weights/sphere with values  
> for the
> newly-interpolated channel?  I think that was one of the concerns that
> Bernhard raised.
>
> J
>
> -- 
> Julie Onton, PhD
> http://sccn.ucsd.edu/~julie
>
>> Dear Bernhard,
>>
>>
>>> As I need to use the algorithm eeg_interp(EEG, Channel, 'spherical')
>>> for
>>> channel-interpolation in combination with ICA I found a possible
>>> contradiction regarding the question what to apply first, ICA or
>>> interpolation.
>>
>>
>> Apply ICA first. Once you have run ICA you may (1) remove artifactual
>> ICA components if this is what you want to do (2) then interpolate
>> channels.
>>
>> If you interpolate channels first, the rank of the data might not be
>> equal to the number of channel. Even if it is (because spherical
>> interpolation is not linear interpolation), you have introduced some
>> non-linearity in the data (in the sense that an underlying source
>> might not project linearly to scalp channels any more) and this is
>> going to make the ICA solution more noisy.
>>
>> Best,
>>
>> Arno
>>
>> _______________________________________________
>> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
>> To unsubscribe, send an empty email to eeglablist-unsubscribe at sccn.ucsd.edu
>> For digest mode, send an email with the subject "set digest mime" to
>> eeglablist-request at sccn.ucsd.edu
>>
>




More information about the eeglablist mailing list