Moving the field forward: detection of epileptiform abnormalities on scalp electroencephalography using deep learning—clinical application perspectives

Abstract

The application of deep learning approaches for the detection of inter-ictal epileptiform discharges is a nascent field, with most studies published in the past 5 years. Although many recent models have been published demonstrating promising results, deficiencies in descriptions of datasets, unstandardized methods, variation in performance evaluation and lack of demonstrable generalizability have made it difficult for these algorithms to be compared and progress to clinical validity. A few recent publications have provided detailed breakdown of datasets and relevant performance metrics to exemplify the potential of deep learning in epileptiform discharge detection. This review provides an overview of the field and equips computer and data scientists with a synopsis of EEG datasets, background and epileptiform variation, model evaluation parameters, and an awareness of the performance metrics of high impact and interest to the trained clinical and neuroscientist EEG end-user. The gold standard and inter-rater disagreements in defining epileptiform abnormalities remain a challenge in the field, and a hierarchical proposal for epileptiform discharge labelling options is recommended. Standardized descriptions of datasets and reporting metrics are a priority. Source code sharing and accessibility to public EEG datasets will increase the rigor, quality and progress in the field and allow validation and real-world clinical translation.

Automated Interpretation of Clinical Electroencephalograms Using Artificial Intelligence

“…To date, AI approaches in clinical EEG have addressed only limited aspects in isolation, such as distinguishing normal from abnormal recordings, detecting seizures, or detecting interictal epileptiform discharges . Other publications have also claimed that AI achieved human expert performance for spike detectors but not for the comprehensive assessment of routine clinical EEGs, equivalent to human expert assessment, which has not yet been reported. Most previously published approaches bear important limitations that are often encountered in AI studies .…”