Visual seizure annotation and automated seizure detection using behind‐the‐ear electroencephalographic channels

Abstract

Objective: Seizure diaries kept by patients are unreliable. Automated electroencephalography (EEG)-based seizure detection systems are a useful support tool to objectively detect and register seizures during long-term video-EEG recording. However, this standard full scalp-EEG recording setup is of limited use outside the hospital, and a discreet, wearable device is needed for capturing seizures in the home setting.We are developing a wearable device that records EEG with behind-the-ear electrodes. In this study, we determined whether the recognition of ictal patterns using only behind-the-ear EEG channels is possible. Second, an automated seizure detection algorithm was developed using only those behind-the-ear EEG channels. Methods: Fifty-four patients with a total of 182 seizures, mostly temporal lobe epilepsy (TLE), and 5284 hours of data, were recorded with a standard video-EEG at University Hospital Leuven. In addition, extra behind-the-ear EEG channels were recorded. First, a neurologist was asked to annotate behind-the-ear EEG segments containing selected seizure and nonseizure fragments. Second, a data-driven algorithm was developed using only behind-the-ear EEG. This algorithm was trained using data from other patients (patient-independent model) or from the same patient (patient-specific model). Results:The visual recognition study resulted in 65.7% sensitivity and 94.4% specificity. By using those seizure annotations, the automated algorithm obtained 64.1% sensitivity and 2.8 false-positive detections (FPs)/24 hours with the patientindependent model. The patient-specific model achieved 69.1% sensitivity and 0.49 FPs/24 hours. Significance: Visual recognition of ictal EEG patterns using only behind-the-ear EEG is possible in a significant number of patients with TLE. A patient-specific seizure detection algorithm using only behind-the-ear EEG was able to detect more seizures automatically than what patients typically report, with 0.49 FPs/24 hours.

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

“…The minimum recording time is 0.74 days and the maximum recording time is 4.9 days. We refer to [23] for a detailed discussion of the content of the dataset. The investigated dataset is from our SeizeIT1 study [19,20].…”

“…Behind-the-ear sensors are able to detect epileptic seizures for focal onset and generalised seizure types [21,22]. In contrast to full scalp EEG, there are only a few studies that investigated automated seizure detection algorithms on behind-the-ear EEG [22][23][24].…”

“…In this article, we defer EEG segments with a length of at least 5 min to a human annotator. We investigated the support vector machine (SVM) models that were trained to perform automated seizure detection on patients with focal epilepsy in [23]. We used the confidences derived from the SVM output or from so-called trust scores [30] (which only depend on the labels and on the features derived from the EEG to train the SVM).…”

“…We used the confidences derived from the SVM output or from so-called trust scores [30] (which only depend on the labels and on the features derived from the EEG to train the SVM). Our second contribution demonstrates that a common modelling strategy, where predictions from several short segments are combined to obtain a final prediction [9,22,23,35], can be improved by filtering out untrustworthy segments. This filtering approach only works with the trust scores.…”

Wearables bei Epilepsien

“…Es gibt mittlerweile auch mobile EEG-Systeme mit auf die Kopfhaut zu platzierenden Trockenelektroden in unterschiedlicher Anzahl, mit denen diagnostisch verwertbare Messungen möglich sind [47], allerdings möchten Patienten keine am Kopf sichtbaren Elektroden tragen [5]. In einer Studie wurden jeweils 2 Oberflächen-Elektroden auf die Kopfhaut hinter das Ohr platziert; bei 182 überwiegend fokalen Anfällen von 54 Patienten (v. a. mit Schläfenlappenepilepsie) wurden mittels automatisierter Detektion eine Sensitivität von 64 % und eine Falschpositiv-Rate von 2,8 pro 24 Stunden in dieser Konstellation erreicht [48]. EEG-Ableitungen im äußeren Gehörgang mit 4 EEG-Elektroden in 15 Patienten mit Schläfenlappenepilepsie lieferten gleich gute Resultate wie simultan aufgezeichnete konventionelle Oberflächen-EEG Registrierun-gen [49].…”