Search

Objective In this study on the analysis of EEG signals for seizure prediction, we used a combination of statistically relevant theory and nonlinear dynamics to maximize the sensitivity of nonlinear analysis and improved prediction accuracy (PA) and effectiveness. Methods First, a physiological reference range of approximate entropy (ApEn) was set up based on normal EEG data. Second, using the concept of global optimization, all EEG electrodes were used in the study regardless of the location of epileptic foci, and the five-electrode group with the strongest synchronization discharge was employed as the optimal electrode group for the next prediction. We set a warning signal when the ApEn values of the data were below the reference range in five electrodes at the same time. Results From the overall 142.7-h EEG signal containing 37 seizures from nine epileptics, our PA was 94.59%, the false prediction rate was 0.084/h, and the mean prediction time was 26.64 min. Conclusion Combining statistically relevant theory and nonlinear dynamics can significantly improve the sensitivity of the nonlinear analysis in seizure prediction. Significance This method may provide a theoretical foundation for the development of a clinical real-time warning system for patients with partial epilepsy. [ABSTRACT FROM AUTHOR]

Abstract: Voltage-gated sodium channels (VGSC) are important determinants of neuronal excitability which are implicated in the pathogenesis of epilepsy. Ankyrin-G contributes to the distribution and regulation of VGSC. Here we investigated the alterations of the two α-subunits SCN8A and SCN1A and their adapter ankyrin-G in the hippocampal cornu ammonis 1 (CA1) of rats after pilocarpine induced status epilepticus (PISE), compared to the sham-control group (C1) and blank-control group (C2). Significant increase of SCN8A mRNA (41.08% increase compared to C1, P <0.001; 30.88% increase compared to C2, P =0.011) was detected 60 days after PISE. At D1 SCN8A mRNA reduced but no significant changes were detected when compared to controls (one-way ANOVA, F =1.232, P =0.276). After measuring the optical density of Western blot, we detected significant differences between the levels of SCN8A protein in different groups but no difference between the protein levels of SCN1A at D1 and D60 after pilocarpine treatment compared to the control. At D60 the relative copies of ankyrin-G mRNA on internal control β-actin in PISE group increased significantly compared to C1 and C2 (one-way ANOVA, F =16.537, P <0.001). Significantly increase of ankyrin-G immunoreactivity in Western blot from the PISE group 1 day and 60 days after PISE was observed, compared to the controls (one-way ANOVA, F =24.255 at D1, P <0.001; F =29.280 at D60, P <0.001). After analyzing the double-stained cells counting, we detected significant differences between the numbers of SCN8A+/ankyrin-G+ immunoreactive cells in different groups in acute and chronic period following PISE (two way-ANOVA, F group =37.905, P <0.001; F day =45.310, P <0.001). The data revealed that both SCN8A and ankyrin-G increased significantly in the CA1 subfield of the rat hippocampus 60 days following pilocarpine induced status epilepticus and co-localized with each other. [Copyright &y& Elsevier]

• The sensitivity of small sharp spikes (SSS) for epilepsy was 18.6 % and the specificity was 97.5 %. • The incidence of SSS was strongly correlated with the frequency of interictal epileptiform discharges (IEDs). • SSS are closely co-lateralized with IEDs in patients with temporal lobe epilepsy (TLE). • SSS can be epileptiform EEG markers in patients with TLE. To determine the incidence and clinical significance of small sharp spikes (SSS) in the patient population of the adult Epilepsy Monitoring Unit (EMU). This is a retrospective study of EEG data and medical records from consecutive patients who underwent video-EEG recording in the adult EMU from March 2013 to February 2019. SSS, interictal epileptiform discharges (IEDs), and ictal patterns were identified. Of the 909 patients reviewed, SSS were observed in110 (12.1 %) patients. Epilepsy was present in 101 of the 110 (91.8 %) patients with SSS and in 441 of the 799 (55.2 %) patients without SSS. The incidence of epilepsy was significantly higher in patients with SSS than in those without SSS (OR = 9.1, 95 % CI: 4.5–18.3, P < 0.01). The sensitivity of SSS for epilepsy was 18.6 % and the specificity was 97.5 %. The incidence of SSS was strongly correlated with the frequency of IEDs (OR 1.89; 95 %CI: 1.60–2.24, P < 0.01). When both present, SSS and IEDs were co-lateralized in the same hemisphere. There is a statistically significant association between SSS and focal epilepsy. SSS have similar clinical implications to IEDs in the lateralization and localization of temporal lobe seizures. SSS can be an epileptiform EEG pattern for temporal lobe epilepsy. [ABSTRACT FROM AUTHOR]

• This paper presents an effective method to preprocess the original EEG data into suitable shape to preserve the temporal and spatial information as much as possible. • A feasible method, long-term recurrent convolutional network (LRCN), was proposed to create epileptic seizure prediction model, which could classify epileptic stages and set an early warning in advance. • The new network structure and instructive settings in the network were proposed. • The post-processing strategy was proposed to predict the seizures in different time scales. The segments-evaluation and events-evaluation were processed to evaluate the performance. • Different methods were made comparisons using the same data in this paper. overall, the LRCN network offers an about 5–9% increase in sensitivity and specificity. A seizure prediction system can detect seizures prior to their occurrence and allow clinicians to provide timely treatment for patients with epilepsy. Research on seizure prediction has progressed from signal processing analyses to machine learning. However, most prediction methods are hand-engineered and have high computational complexity, increasing the difficulty of obtaining real-time predictions. Some forecasting and early warning methods have achieved good results in the short term but have low applicability in practical situations over the long term. First, electroencephalogram (EEG) time series were converted into two-dimensional images for multichannel fusion. A feasible method, a long-term recurrent convolutional network (LRCN), was proposed to create a spatiotemporal deep learning model for predicting epileptic seizures. The convolutional network block was used to automatically extract deep features from the data. The long short-term memory (LSTM) block was incorporated into learning a time sequence for identifying the preictal segments. New network settings and a postprocessing strategy were proposed in the seizure prediction model. The deep seizure prediction model achieved an accuracy of 93.40%, prediction sensitivity of 91.88% and specificity of 86.13% in segment-based evaluations. For the event-based evaluations, 164 seizures were predicted. The proposed method provides high sensitivity and a low false prediction rate (FPR) of 0.04 F P/h. We employed different methods, including the LRCN, deep learning and traditional machine learning methods, and compared them using the same data in this paper. Overall, the LRCN offers approximately 5–9% increased sensitivity and specificity. This study describes the LRCN network for analyzing EEG data to predict epileptic seizures, thereby enabling the implementation of an early warning system that detects epileptic seizures before they occur in clinical applications. [ABSTRACT FROM AUTHOR]