Computer-assisted EEG diagnostic review for idiopathic generalized epilepsy

Abstract

Epilepsy diagnosis can be costly, time-consuming and not uncommonly inaccurate. The reference standard diagnostic monitoring is continuous video-EEG monitoring, ideally capturing all events or concordant interictal discharges. Automating EEG data review would save time and resources, thus enabling more people to receive reference standard monitoring and also potentially herald a more quantitative approach to therapeutic outcomes. There is substantial research into automated detection of seizures and epileptic activity from EEG. However, automated detection software is not widely used in the clinic; and, despite numerous published algorithms, few methods have regulatory approval for detecting epileptic activity from EEG.This study reports on a deep learning algorithm for computer-assisted EEG review. Deep, convolutional neural networks were trained to detect epileptic discharges using a pre-existing dataset of over 6000 labelled events in a cohort of 103 patients with idiopathic generalized epilepsy (IGE). Patients underwent 24-hour ambulatory outpatient EEG, and all data was curated and confirmed independently by two epilepsy specialists . The resulting automated detection algorithm was then used to review diagnostic scalp-EEG for seven patients (four with IGE and three with events mimicking seizures) to validate performance in a clinical setting.The automated detection algorithm showed state-of-the-art performance for detecting epileptic activity from clinical EEG, with mean sensitivity of >95% and corresponding mean false positive rate of 1 detection per minute. Importantly, diagnostic case studies showed that the automated detection algorithm reduced human review time by 80%-99%, without .

Machine learning and wearable devices of the future

“…With an algorithm developed using ML, the multimodal WD detected GTCS with high sensitivity (92%-100%) and low false alarm rate (0.2-1/day), in phase 2 studies. 42 In the diagnostic space, ML approaches can expedite clinical review of diagnostic scalp EEG, [43][44][45] although currently there are relatively few approved algorithms for automated EEG review. 14 ML has also been extensively applied to the problem of seizure forecasting.…”

Past, Present and Future of Home video‐electroencephalographic telemetry: A review of the development of in‐home video‐electroencephalographic recordings

“…The cloud environment enables collaboration, curation, and new research, such as the development of machine tools. 15…”

“…For example, we found that nonepileptic seizures had a significantly shorter duration at home compared to the hospital setting (mean = 320 seconds at home vs mean = 868 seconds in the hospital, unpublished data from King's College Hospital). The cloud environment enables collaboration, curation, and new research, such as the development of machine tools 15 Current challenges and limitations …”

Classification with a Deferral Option and Low-Trust Filtering for Automated Seizure Detection

“…Our first contribution in this paper, is the evaluation of the performance of the classifier in case that the EEG segments for which the classifier is least confident are deferred to a human annotator, who is assumed to annotate perfectly. A similar scenario is quite common in clinical epilepsy research: the algorithm flags all suspicious activity, which is then presented to a human annotator [25]. Learning algorithms with a reject option have a long history in machine learning research [26,27].…”

“…Application of this approach to seizure detection has been limited. Computer-assisted detection of epileptic discharges from full scalp EEG has been investigated by Clarke et al [25]. On a retrospective dataset, a neural network achieved a DS of 96.7% with 1670 false detections per 24 h. They employed this model in a clinical application of ambulatory measurement of 7 patients with idiopathic generalised epilepsy.…”