RESUMO
Background: Out-of-hospital seizure detection aims to provide clinicians and patients with objective seizure documentation in efforts to improve the clinical management of epilepsy. In-patient studies have found that combining different modalities helps improve the seizure detection accuracy. In this study, the objective was to evaluate the viability of out-of-hospital seizure detection using wearable ECG, accelerometry and behind-the-ear electroencephalography (EEG). Furthermore, we examined the signal quality of out-of-hospital EEG recordings. Methods: Seventeen patients were monitored for up to 5 days. A support vector machine based seizure detection algorithm was applied using both in-patient seizures and out-of-hospital electrographic seizures in one patient. To assess the content of noise in the EEG signal, we compared the root-mean-square (RMS) of the recordings to a reference threshold derived from manually categorised segments of EEG recordings. Results: In total 1427 hours of continuous EEG was recorded. In one patient, we identified 15 electrographic focal impaired awareness seizures with a motor component. After training our algorithm on in-patient data, we found a sensitivity of 91% and a false alarm rate (FAR) of 18/24 hours for the detection of out-of-hospital seizures using a combination of EEG and ECG recordings. We estimated that 30.1% of the recorded EEG signal was physiological EEG, with an RMS value within the reference threshold. Conclusion: We found that detection of out-of-hospital focal impaired awareness seizures with a motor component is possible and that applying multiple modalities improves the diagnostic accuracy compared with unimodal EEG. However, significant challenges remain regarding a high FAR and that only 30.1% of the EEG data represented usable signal.
RESUMO
BACKGROUND: We propose rigorously optimised supervised feature extraction methods for multilinear data based on Multilinear Discriminant Analysis (MDA) and demonstrate their usage on Electroencephalography (EEG) and simulated data. While existing MDA methods use heuristic optimisation procedures based on an ambiguous Tucker structure, we propose a rigorous approach via optimisation on the cross-product of Stiefel manifolds. We also introduce MDA methods with the PARAFAC structure. We compare the proposed approaches to existing MDA methods and unsupervised multilinear decompositions. RESULTS: We find that manifold optimisation substantially improves MDA objective functions relative to existing methods and on simulated data in general improve classification performance. However, we find similar classification performance when applied to the electroencephalography data. Furthermore, supervised approaches substantially outperform unsupervised mulitilinear methods whereas methods with the PARAFAC structure perform similarly to those with Tucker structures. Notably, despite applying the MDA procedures to raw Brain-Computer Interface data, their performances are on par with results employing ample pre-processing and they extract discriminatory patterns similar to the brain activity known to be elicited in the investigated EEG paradigms. CONCLUSION: The proposed usage of manifold optimisation constitutes the first rigorous and monotonous optimisation approach for MDA methods and allows for MDA with the PARAFAC structure. Our results show that MDA methods applied to raw EEG data can extract discriminatory patterns when compared to traditional unsupervised multilinear feature extraction approaches, whereas the proposed PARAFAC structured MDA models provide meaningful patterns of activity.
