Systems, Subjects, Sessions: To What Extent Do These Factors Influence EEG Data?

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Electroencephalography (EEG) is a mature neuroimaging technique producing high-quality scientific data.
• The impact of different EEG system choices on data variance is often underestimated.
• The proliferation of low-cost and mobile EEG systems necessitates an evaluation of their data consistency.

Purpose of the Study:

• To quantify and compare the variance introduced by different EEG systems, subjects, and repeated sessions.
• To assess the reliability of data from mobile and research-grade EEG systems.
• To propose a benchmark for evaluating new mobile EEG systems using event-related potentials (ERPs).

Main Methods:

• Four EEG systems were tested: two research-grade and two mobile systems (one dry-electrode, one low-channel count).
• Four subjects were recorded three times with each system, performing six standard EEG paradigms.
• Variance was analyzed across systems, subjects, and sessions for event-related potentials (ERPs) and steady-state visually evoked potentials (SSVEPs).

Main Results:

• Subjects accounted for 32% of the variance, EEG systems for 9%, and repeated sessions for 1%.
• With 16 subjects, system variance (9%) is comparable to subject variance (approx. 8%).
• Mobile EEG systems showed significantly different mean values from research-grade systems in several paradigms.

Conclusions:

• EEG system choice is a significant source of data variance, comparable to subject variability in studies with sufficient participants.
• The reliability of data from mobile EEG systems needs careful consideration, as they can introduce systematic differences.
• A benchmark using ERP responses is proposed for evaluating the performance and consistency of new mobile EEG devices.