We use a 128 electrode array, but I would also be curious about 64. I am less curious about lower density arrays since bad electrodes should be relatively rare. I am also guessing the type of analyses being run are significant factors (ERP, Time-Frequency, etc.). Thank you!
It definitely depends on the number of electrodes, their locations on the scalp or face and the analysis method but in my work with the EGI 128 nets I used a rule of rejecting epochs with more than 20 bad channels / 15% of electrodes. I worked with children and rarely hit this threshold when the EEG net was applied correctly. We trained our data collection RAs to continuously review data quality during acquisition and adjust any electrodes, particularly face / eye electrodes or those in our montage of interest. While density improves your ability to interpolate bad electrodes, it's always better to fix bad electrodes.
Agreeing with Trey Avery it really does depend on several factors. We’ve used both 124 and 64 Channel EGI Hyrodrocel GSN 30 nets. when Testing we try to make sure that the area surrounding the center of the reference, and the point of reference, have impedances below 50 mV. However it’s my understanding that impedances to 100mV are acceptable and have resulted in good data during analysis.
Throughout testing we try to check the impedances during breaks. We re-wet electrodes with high impedance and ensure their all making contact with the scalp (or the face).
I agree with Trey Avery . It really depends on a number of factors, but I think the 15% threshold is a good rule of thumb. However, it depends on what you're doing. For instance, if you are only planning to estimate event-related potentials, and none of the bad channels are relevant for that particular ERP, you may end up accepting more bad channels to interpolate.
- Badges
- Science topic