Assessing the Quality of Wearable EEG Systems Using Functional Connectivity

Assessing the data quality of wearable electroencephalogram (EEG) systems is critical to collecting reliable neurophysiological data in non-laboratory environments. To date, measures of signal quality and spectral characteristics have been used to characterize wearable EEG systems. We demonstrate that these traditional measures do not provide fine-grained differentiation between the performance of four popular wearable EEG systems (the Epoc+, OpenBCI, DSI-24 and Quick-30 Dry EEG). Using two computationally inexpensive metrics of undirected functional connectivity (phase lag index) and directed functional connectivity (directed phase lag index), we compare the integrity of the phase relationships captured by wearable systems to those recorded from a high-density research-grade EEG system (Electrical Geodesics Inc). Our results demonstrate that functional connectivity analyses provide additional discriminatory information about wearable EEG systems, with clear differentiation of the cosine similarity between research-grade functional connectivity patterns and those generated by each wearable system. We provide a freely available Matlab toolbox containing all metrics described in this paper such that researchers and non-experts interested in wearable EEG systems can easily assess the quality of systems not characterized in this study, thus advancing the translation of EEG research into non-laboratory settings.