[FieldTrip] Artifact rejection on average ERF

[Judy Zhu]

Dear Fieldtrippers,

I am attempting to remove some artifact in my data using ICA component
rejection. I would like your advice on whether it would be valid to perform
artifact rejection on the averaged ERF (i.e. timelock output), and how this
may affect the covariance matrix. I have tried doing the artifact rejection
on all individual epochs before computing ERF, but this hasn't worked
consistently well across subjects, therefore I would like to try it on the
average ERFs for each subject.

I am doing the following steps:
- compute the average ERF (for each condition) and the covariance matrix
(based on all conditions combined together) using ft_timelockanalysis
- perform artifact rejection on the average ERFs using ft_rejectcomponent
- use the cleaned ERFs for lcmv beamforming

My questions is:
Can I use the cleaned ERFs with the previously-computed covariance matrix
to do beamforming? (or is that covariance matrix now invalid?) Is there a
way I can re-compute the covariance matrix for the cleaned ERFs?

Many thanks for your help in advance!

Thanks,
Judy


Judy D. Zhu
PhD Candidate
ARC Centre of Excellence in Cognition and its Disorders
Department of Cognitive Science, Macquarie University, Sydney, Australia
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20180917/e5523e55/attachment.html>