Abstract
Among diagnostic biomarkers, high frequency oscillations in human iEEG are used to identify epileptogenic brain tissue during epilepsy surgery. However, current methods typically analyse the raw data offline using complex time-consuming algorithms. We developed a compact neuromorphic sensory-processing system-on-chip that can monitor the iEEG signals and detect high frequency oscillations in real-time using spiking neural networks. To this end, we present an integrated device with an analog front-end that can extract predefined spectral features and encode them as address-events, and a neuromorphic processor core that implements a network of integrate and fire neurons with dynamic synapses. The device was fabricated using a standard 0.18μm CMOS technology node. The estimated power consumption of the front-end is 6.2μW /channel and the area-on-chip for a single channel is 0.15 square millimetres. The SNN classifier provides 90.5% sensitivity and 67.7% specificity for detecting high frequency oscillations. This is the first feasibility study towards identifying relevant features in intracranial human data in real-time on-chip using event-base processors.
Sharifshazileh, Mohammadali