Systematic design and implementation of a high-robust adaptive calibration technique for ETI-induced analog front end circuits in EEG systems

Abstract

Robustness is a major concern of wearable electroencephalogram (EEG) acquisition systems. It is often closely tied with variations of electrode to tissue interface (ETI), resulting in significant difficulties for following analog front end (AFE) circuitry design and signal processing. To realize a high-robust EEG system and relax the AFE requirements, adaptive ETI characterization and signal calibration techniques are proposed. Instead of monitoring ETI impedance and/or employing an active electrode, this paper demonstrates an attempt to continuously detect dc offset and ac-coupled gain variations of an entire ETI-induced signal acquisition path (ETI-AP). Both additional and multiplicative factors of distorted signals are stabilized by an adaptive coefficient compensation algorithm, which modulates the non-stationary ETI-AP to a constant channel. During EEG measurement, a pseudo-random number sequence generated from a FPGA is driven to body as a test signal to estimate and compensate the ETI-AP characteristic. The proposed techniques were evaluated in lab environments where both spontaneous and evoked EEGs were recorded by using a bipolar montage on a long-haired, healthy adult. Furthermore, foam-covered and pin-based electrodes were placed on the participant's scalp and skin surface in order to establish different ETI conditions. With this proposed technique, the empirical results shown the dynamic variation of non-stationary ETI was continuously characterized; calibrated EEGs achieved higher signal to noise ratio, which demonstrates the validity of the proposed method, as well as its compatibility with diverse sensors and bio-medical acquisition systems.