Dear Jay,
Not sure what you mean by PCA but ICA is often used for removing
artefacts, particularly eye blinks and eye movements. Whether this
will make a difference depends on how much of those artefacts are
present in your data and what kind of analysis you do so there is no
way to give a generic answer. To give you extreme examples: if your
subjects was instructed to fixate and not to blink during experimental
trials and there were no blinks at all then ICA will not make a
difference. If the subject blinked consistently at a particular time
in the trial and you look at ERP, the ERP will be dominated by
eyeblinks and to this kind of data ICA would make a huge difference. I
don't know about papers that discuss specifically this but there
probably are some.
Vladimir
On Tue, Sep 25, 2012 at 3:39 PM, jay chen <[log in to unmask]> wrote:
> Dear Dr. Litvak:
>
> To reduce noise, some researchers use PCA and ICA for EEG
> signal processing. But I wonder how much quantitative difference in
> EEG signal between with and without PCA and ICA steps? Are there
> other reasons that we need to perform PCA and ICA steps? Any papers
> discuss this issue? I appreciate your help.
>
>
> Best Regards
> Jay Chen
|