Automated Spectrographic Seizure Detection Using Convolutional Neural Networks

Non-convulsive seizures (NCS) are common in critically ill patients (8-50%) [1–6], and if untreated, are associated with high mortality (17-51%) [6–8]. Furthermore, effective treatment must be timely, as delayed treatment can lead to long term neurologic disability [9–11]. Many intensive care units (ICU) now perform continuous EEGs (cEEG) to monitor for NCS. However, although continuously recorded, cEEGs are rarely continuously monitored. Even in large academic medical centers, cEEGs are typically reviewed 2-3 times daily by a neurophysiologist.

0