Automated identification and localization of interictal epileptiform discharges: leveraging morphological analysis, five-criterion fulfillment, and machine learning approach

Interictal epileptiform discharges (IEDs) play a crucial role in the diagnosis and assessment of seizure risk in epilepsy. Their frequency, amplitude, and morphological characteristics serve as consistent markers of epileptogenesis. Currently, the clinical evaluation of IEDs heavily relies on visual detection by specialized experts, which is a subjective and time-consuming task. To address this, there is a need for an automated IEDs detection system that can provide faster and more reliable epilepsy diagnosis. In this paper, we propose a novel method for automatic identification of IEDs by inspecting the fulfillment of specific criteria. The decisions made by our method were compared with the combined decisions of three neurologists, serving as a benchmark for evaluation. The results and performance metrics obtained from this comparison demonstrate the effectiveness and potential of our automated IEDs identification method.