Modular Flexible 80-dB-DR Artifact-Resilient EEG Headset with Distributed Pulse-Based Feature Extraction and Multiplier-Less Neuromorphic Boosted Seizure Classifier

Wearable EEG headsets have shown potential to transform outpatient diagnostics by providing real-time insights into brain neurological activity, allowing for more accurate treatment plans. For most diagnostic applications, energy-efficient design is crucial due to the need for long-term recording. Diagnostic headsets typically consist of multiple active electrodes (AE) with embedded electronics for amplification and/or quantization, connected to a central back-end (BE) unit responsible for data processing and, if necessary, wireless transmission. As shown in Fig. 1 (top, left), a review of the state of the art reveals that in systems with a sufficiently-high dynamic range (DR) analog front-end (AFE) [1] and a data-driven classifier (e.g., a nonlinear support vector machine (NL-SVM) [2]) for seizure detection, power consumption is mainly dominated by the AFE (47.6%), AE-to-BE data communication (26.5%), and signal processing for seizure detection (20.6^%). This emphasizes the need for a holistic approach to enhance the efficiency of all these major components for an overall energy-efficient design.