Automated Inter-Ictal Epileptiform Discharge Detection from Routine EEG

Epilepsy is the most common neurological disorder. The diagnosis commonly requires manual visual electroencephalogram (EEG) analysis which time-consuming. Deep learning has shown promising performance in detecting interictal epileptiform discharges (IED) and may improve quality of epilepsy monitoring. However, datasets literature are small (n≤100) collected from single clinical centre, limiting generalization across different devices settings. To better automate IED detection, we cross-evaluated a Resnet architecture on 2 sets routine EEG recordings patients with idiopathic generalized at Alfred Health Hospital Royal Melbourne (RMH). We split these into 2s windows or without evaluated model variants terms how well they classified windows. results our experiment showed that an AUC score 0.894 (95% CI, 0.881–0.907) when trained Alfred’s dataset tested RMH’s dataset, 0.857 0.847–0.867) vice versa. In addition, compared best variant Persyst observed was comparable.