Neuromorphic deep spiking neural networks for seizure detection

Abstract The vast majority of studies that process and analyze neural signals are conducted on cloud computing resources, which is often necessary for the demanding requirements deep network workloads. However, applications such as epileptic seizure detection stand to benefit from edge devices can securely sensitive medical data in a real-time personalised manner. In this work, we propose novel neuromorphic approach using surrogate gradient-based spiking (SNN), consists ConvLSTM unit. We have trained, validated, rigorously tested proposed SNN model across three publicly accessible datasets, including Boston Children’s Hospital–MIT (CHB-MIT) dataset U.S., Freiburg (FB) EPILEPSIAE intracranial electroencephalogram datasets Germany. average leave-one-out cross-validation area under curve score FB, CHB-MIT reach 92.7 % , 89.0 81.1 respectively, while computational overhead energy consumption significantly reduced when compared alternative state-of-the-art models, showing potential building an accurate hardware-friendly, low-power system. This first feasibility study several reliable public datasets.