Impact of amplitude errors and uncertainties in the measurement of brain potentials through wearable EEG devices

This work investigates the amplitude of signals measured by electroencephalographic devices. Indeed, acquisition latencies may affect the detection of time-locked neural phenomena, especially in wireless equipment. Nonetheless, further sources of errors and uncertainties have a potential impact on the measurement of brain potentials and might consequently lead to wrong conclusions. Therefore, a measurement station was used to generate a reference voltage signal and acquire it simultaneously from a voltmeter and EEG devices under test. Then, data was processed with a grand average of aligned 8 s-long epochs, thus simulating a typical processing pipeline for event related potentials detection. Amplitude errors were retrieved under this condition, and uncertainties were evaluated for two representative wireless devices. Relative amplitude errors were found to be about 14 % in one device and up 45 % for another. Meanwhile, relative uncertainties were typically in the 28 % to 61 % range with even a maximum of 167 % in one case. These results are discussed in relation to an experiment with a brain-computer interface, and suggest that such errors and uncertainties can significantly impact the detection of inherent brain potentials.