Good data? The EEG Quality Index for Automated Assessment of Signal Quality

Translating EEG technology into real world applications is limited by a variety of challenges, one of which is signal noise from portable, low-density EEG devices operating in uncontrolled environments. We have developed an automated general-purpose framework, the EEG Quality Index, for the objective assessment of EEG signal quality. This framework currently uses 6 metrics that are shown to be highly sensitive to various typical artifacts. Normalized scores for each metric can be used in a framework to provide indications of the overall quality of the recorded EEG signal. Using simulated clean and artifact-contaminated EEG in a Support Vector Machine classifier, the EEG Quality Index metrics can identify artifact-contaminated EEG with 90.75% accuracy, 97.7% sensitivity, and 83.8% specificity. This is the first demonstration of a general-purpose framework for the automated and objective assessment of the quality of EEG data. With portable, low-density EEG systems being increasingly adopted in real-world clinical applications, this type of quality assessment is important.