[FieldTrip] Automatic artifact rejection settings

[Helen L. Wieffering]

Hi all,

I have a question related to the z-value cutoffs and filter settings in the ft_artifact_zvalue function. After running through the 'default' settings suggested in the tutorial on our own EEG datasets, we've been able to consistently identify EOG artifacts but have had less success when it comes to muscle/jump settings. Often the function is oversensitive and identifies far too many artifacts than we could possibly remove, while occasionally still failing to mark substantial drifts or jumps.

I realize each dataset is different and there are rarely 'one size fits all' z-value settings, but I'd appreciate hearing if anyone out there has had similar difficulty and/or later success in tailoring the settings to accurately mark their own data. There's very little explanation given as to how the particular values in the tutorial were chosen or how to go about changing them ... e.g., what is a median filter, and why it it best suited for detecting jump artifacts? I'd love to know more background on those parameters if possible.

Thanks for any help!

Best,

Helen Wieffering
Bowdoin College


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20150714/b0388345/attachment-0001.html>