Peri-ictal heart rate variability parameters as surrogate markers of seizure severity.

This study aims at defining objective parameters reflecting the severity of peri-ictal autonomic changes and their relation to post-ictal generalized electroencephalography (EEG) suppression (PGES), with the view that such changes could be detected by wearable seizure detection systems and prove useful to assess the risk of sudden unexpected death in epilepsy (SUDEP). To this purpose, we assessed peri-ictal changes in heart rate variability (HRV) and correlated them with seizure duration, intensity of electromyography-based ictal muscle activity, and presence and duration of post-ictal generalized EEG suppression (PGES). We evaluated 75 motor seizures from 40 patients, including 61 generalized tonic-clonic seizures (GTCS) and 14 other major motor seizure types. For all major motor seizures, HRV measurements demonstrated a significantly decreased parasympathetic activity and increased sympathetic activity in the post-ictal period. The post-ictal increased sympathetic activity was significantly higher for GTCS as compared with non-GTCS. The degree of peri-ictal decreased parasympathetic activity and increased sympathetic activity was associated with longer PGES (>20 s), longer seizure duration, and greater intensity of ictal muscle activity. Mean post-ictal heart rate (HR) was an independent predictor of PGES duration, seizure duration, and intensity of ictal muscle contraction. Our results indicate that peri-ictal changes in HRV are potential biomarkers of major motor seizure severity.

Detection of convulsive seizures using surface electromyography.

Bilateral (generalized) tonic-clonic seizures (TCS) increase the risk of sudden unexpected death in epilepsy (SUDEP), especially when patients are unattended. In sleep, TCS often remain unnoticed, which can result in suboptimal treatment decisions. There is a need for automated detection of these major epileptic seizures, using wearable devices. Quantitative surface electromyography (EMG) changes are specific for TCS and characterized by a dynamic evolution of low- and high-frequency signal components. Algorithms targeting increase in high-frequency EMG signals constitute biomarkers of TCS; they can be used both for seizure detection and for differentiating TCS from convulsive nonepileptic seizures. Two large-scale, blinded, prospective studies demonstrated the accuracy of wearable EMG devices for detecting TCS with high sensitivity (76%-100%). The rate of false alarms (0.7-2.5/24 h) needs further improvement. This article summarizes the pathophysiology of muscle activation during convulsive seizures and reviews the published evidence on the accuracy of EMG-based seizure detection.

Automated differentiation between epileptic and nonepileptic convulsive seizures.

Our objective was the clinical validation of an automated algorithm based on surface electromyography (EMG) for differentiation between convulsive epileptic and psychogenic nonepileptic seizures (PNESs). Forty-four consecutive episodes with convulsive events were automatically analyzed with the algorithm: 25 generalized tonic-clonic seizures (GTCSs) from 11 patients, and 19 episodes of convulsive PNES from 13 patients. The gold standard was the interpretation of the video-electroencephalographic recordings by experts blinded to the EMG results. The algorithm correctly classified 24 GTCSs (96%) and 18 PNESs (95%). The overall diagnostic accuracy was 95%. This algorithm is useful for distinguishing between epileptic and psychogenic convulsive seizures.

Quantitative analysis of surface electromyography during epileptic and nonepileptic convulsive seizures.

OBJECTIVE: To investigate the characteristics of sustained muscle activation during convulsive epileptic and psychogenic nonepileptic seizures (PNES), as compared to voluntary muscle activation. The main goal was to find surface electromyography (EMG) features that can distinguish between convulsive epileptic seizures and convulsive PNES. METHODS: In this case-control study, surface EMG was recorded from the deltoid muscles during long-term video-electroencephalography (EEG) monitoring in 25 patients and in 21 healthy controls. A total of 46 clinical episodes were recorded: 28 generalized tonic-clonic seizures (GTCS) from 14 patients with epilepsy, and 18 convulsive PNES from 12 patients (one patient had both GTCS and PNES). The healthy controls were simulating GTCS. To quantitatively characterize the signals we calculated the following parameters: root mean square (RMS) of the amplitude, median frequency (MF), coherence, and duration of the seizures, of the clonic EMG discharges, and of the silent periods between the cloni. Based on wavelet analysis, we distinguished between a low-frequency component (LF 2-8 Hz) and a high-frequency component (HF 64-256 Hz). RESULTS: Duration of the seizure, and separation between the tonic and the clonic phases distinguished at group-level but not at individual level between convulsive PNES and GTCS. RMS, temporal dynamics of the HF/LF ratio, and the evolution of the silent periods differentiated between epileptic and nonepileptic convulsive seizures at the individual level. A combination between HF/LF ratio and RMS separated all PNES from the GTCS. A blinded review of the EMG features distinguished correctly between GTCS and convulsive PNES in all cases. The HF/LF ratio and the RMS of the PNES were smaller compared to the simulated seizures. SIGNIFICANCE: In addition to providing insight into the mechanism of muscle activation during convulsive PNES, these results have diagnostic significance, at the individual level. Surface EMG features can accurately distinguish convulsive epileptic from nonepileptic psychogenic seizures, even in PNES cases without rhythmic clonic movements.

Modulation of epileptiform EEG discharges in juvenile myoclonic epilepsy: an investigation of reflex epileptic traits.

PURPOSE: Previous studies have suggested that cognitive tasks modulate (provoke or inhibit) the epileptiform electroencephalography (EEG) discharges (EDs) in patients with juvenile myoclonic epilepsy (JME). Their inhibitory effect was found to be especially frequent (64-90%). These studies arbitrarily defined modulation as a >100% increase or >50% decrease of the EDs compared with baseline, which may not sufficiently distinguish from spontaneous fluctuations. The aim of our study was to assess the modulation of EDs and the precipitation of myoclonic seizures by cognitive tasks and by conventional provocation methods, taking into account also the spontaneous fluctuation of EDs. METHOD: Sixty patients with JME underwent video-EEG recordings including 50-min baseline, sleep, hyperventilation, intermittent photic stimulation (IPS), and cognitive tasks. To account for spontaneous fluctuations of the EDs we divided the baseline period into 5-min epochs and calculated the 95% confidence interval for the baseline EDs in each patient. Modulation was assumed when the number of EDs during any 5-min test period was outside the 95% confidence interval. KEY FINDINGS: Using the arbitrary method, our results were similar to previous publications: Cognitive tasks seemed to inhibit the EDs in 94% of the patients, and to provoke them in 22%. However, when the spontaneous fluctuations were accounted for, inhibition was found in only 29% of the patients and provocation in 18%. A nonspecific effect of any cognitive task seemed to account for the observed significant inhibition in two-thirds of the cases, but was observed in only one of the patients with significant provocation. Photoparoxysmal response was observed in 23% of the patients. When accounting for the spontaneous occurrence of EDs, IPS had provocative effect in 10% of the patients. Hyperventilation and sleep had provocative effect on EDs to an extent similar to the cognitive tasks (hyperventilation: 22%; sleep: 18%). The conventional provocation methods tended to be more efficient in patients who were not seizure free. Myoclonia were recorded most often during the cognitive tasks (10 patients). SIGNIFICANCE: Spontaneous fluctuations of EDs account for most of the previously described inhibitory effect of the cognitive tasks. The provocative effect of the cognitive tasks is task-specific, whereas the inhibitory effect seems to be related to cognitive activation in general.

Patterns of muscle activation during generalized tonic and tonic-clonic epileptic seizures.

PURPOSE: Tonic seizures and the tonic phase of tonic-clonic epileptic seizures are defined as "sustained tonic" muscle contraction lasting a few seconds to minutes. Visual inspection of the surface electromyogram (EMG) during seizures contributed considerably to a better understanding and accurate diagnosis of several seizure types. However, quantitative analysis of the surface EMG during the epileptic seizures has received surprisingly little attention until now. The aim of our study was to elucidate the pathomechanism of the tonic muscle activation during epileptic seizures. METHODS: Surface EMG was recorded from the deltoid muscles, on both sides, during 63 seizures from 20 patients with epilepsy (10 with generalized tonic and 10 with tonic-clonic seizures). Twenty age- and gender-matched normal controls simulated 100 generalized tonic seizures. To characterize the signal properties we calculated the root mean square (RMS) of the amplitudes, the median frequency (MF), and the coherence. Based on the spectrograms of both epileptic and simulated seizures, we chose to determine the relative spectral power (RP) in the higher (100-500 Hz) frequency domain. KEY FINDINGS: During the tonic seizures there was a significant shift toward higher frequencies, expressed by an increase in the MF and the RP (100-500 Hz). The amplitude characteristic of the signal (RMS) was significantly higher during the tonic phase of the tonic-clonic seizures as compared to the simulated ones, whereas the RMS of the tonic seizures was significantly lower than the simulated ones. The EMG-EMG coherence was significantly higher during the epileptic seizures (both types) as compared to the simulated ones. SIGNIFICANCE: Our results indicate that the mechanism of muscle activation during epileptic seizures is different from the physiologic one. Furthermore the sustained muscle activation during the tonic phase of tonic-clonic seizures is different from that during tonic seizures: The tonic phase of tonic-clonic seizures is characterized by increased amplitude of the signal, whereas tonic seizures are produced by a significant increase in the frequency of the signal.

Ictal quantitative surface electromyography correlates with postictal EEG suppression.

OBJECTIVE: To test the hypothesis that neurophysiologic biomarkers of muscle activation during convulsive seizures reveal seizure severity and to determine whether automatically computed surface EMG parameters during seizures can predict postictal generalized EEG suppression (PGES), indicating increased risk for sudden unexpected death in epilepsy. Wearable EMG devices have been clinically validated for automated detection of generalized tonic-clonic seizures. Our goal was to use quantitative EMG measurements for seizure characterization and risk assessment. METHODS: Quantitative parameters were computed from surface EMGs recorded during convulsive seizures from deltoid and brachial biceps muscles in patients admitted to long-term video-EEG monitoring. Parameters evaluated were the durations of the seizure phases (tonic, clonic), durations of the clonic bursts and silent periods, and the dynamics of their evolution (slope). We compared them with the duration of the PGES. RESULTS: We found significant correlations between quantitative surface EMG parameters and the duration of PGES (p < 0.001). Stepwise multiple regression analysis identified as independent predictors in deltoid muscle the duration of the clonic phase and in biceps muscle the duration of the tonic-clonic phases, the average silent period, and the slopes of the silent period and clonic bursts. The surface EMG-based algorithm identified seizures at increased risk (PGES ≥20 seconds) with an accuracy of 85%. CONCLUSIONS: Ictal quantitative surface EMG parameters correlate with PGES and may identify seizures at high risk. CLASSIFICATION OF EVIDENCE: This study provides Class II evidence that during convulsive seizures, surface EMG parameters are associated with prolonged postictal generalized EEG suppression.

Automated real-time detection of tonic-clonic seizures using a wearable EMG device.

OBJECTIVE: To determine the accuracy of automated detection of generalized tonic-clonic seizures (GTCS) using a wearable surface EMG device. METHODS: We prospectively tested the technical performance and diagnostic accuracy of real-time seizure detection using a wearable surface EMG device. The seizure detection algorithm and the cutoff values were prespecified. A total of 71 patients, referred to long-term video-EEG monitoring, on suspicion of GTCS, were recruited in 3 centers. Seizure detection was real-time and fully automated. The reference standard was the evaluation of video-EEG recordings by trained experts, who were blinded to data from the device. Reading the seizure logs from the device was done blinded to all other data. RESULTS: The mean recording time per patient was 53.18 hours. Total recording time was 3735.5 hours, and device deficiency time was 193 hours (4.9% of the total time the device was turned on). No adverse events occurred. The sensitivity of the wearable device was 93.8% (30 out of 32 GTCS were detected). Median seizure detection latency was 9 seconds (range -4 to 48 seconds). False alarm rate was 0.67/d. CONCLUSIONS: The performance of the wearable EMG device fulfilled the requirements of patients: it detected GTCS with a sensitivity exceeding 90% and detection latency within 30 seconds. CLASSIFICATION OF EVIDENCE: This study provides Class II evidence that for people with a history of GTCS, a wearable EMG device accurately detects GTCS (sensitivity 93.8%, false alarm rate 0.67/d).

Quantitative analysis of surface electromyography: Biomarkers for convulsive seizures.

Muscle activity during seizures is in electroencephalographical (EEG) praxis often considered an irritating artefact. This article discusses ways by surface electromyography (EMG) to turn it into a valuable tool of epileptology. Muscles are in direct synaptic contact with motor neurons. Therefore, EMG signals provide direct information about the electric activity in the motor cortex. Qualitative analysis of EMG has traditionally been a part of the long-term video-EEG recordings. Recent development in quantitative analysis of EMG signals yielded valuable information on the pathomechanisms of convulsive seizures, demonstrating that it was different from maximal voluntary contraction, and different from convulsive psychogenic non-epileptic seizures. Furthermore, the tonic phase of the generalised tonic-clonic seizures (GTCS) proved to have different quantitative features than tonic seizures. The high temporal resolution of EMG allowed detailed characterisation of temporal dynamics of the GTCS, suggesting that the same inhibitory mechanisms that try to prevent the build-up of the seizure activity, contribute to ending the seizure. These findings have clinical implications: the quantitative EMG features provided the pathophysiologic substrate for developing neurophysiologic biomarkers that accurately identify GTCS. This proved to be efficient both for seizure detection and for objective, automated distinction between convulsive and non-convulsive epileptic seizures.

Detection of tonic epileptic seizures based on surface electromyography.

The purpose of this project was to design an algorithm for detection of tonic seizures based on surface electromyography signals from the deltoids. A successful algorithm has a future prospect of being implemented in a wearable device as part of an alarm system. This has already been done for generalized tonic-clonic seizures, and the hypothesis was that some of the same characteristics could be found for tonic seizures. The signals were pre-processed by a high-pass filter to remove low frequency noise such as movement artifacts. Several different features were investigated, including kurtosis, median frequency, zero crossing rate and approximate entropy. These features were used as input in the random forest classifier to decide if a data segment was from a seizure or not. The goal was to develop a generic algorithm for all tonic seizures, but better results were achieved when certain parameters were adapted specifically for each patient. With patient specific parameters the algorithm obtained a sensitivity of 100% for four of six patients with false detection rates between 0.08 and 7.90 per hour.

Dynamics of muscle activation during tonic-clonic seizures.

The purpose of our study was to elucidate the dynamics of muscle activation during generalised tonic-clonic seizures (GTCS). We recorded surface electromyography (EMG) from the deltoid muscle during 26 GTCS from 13 patients and compared it with GTCS-like events acted by 10 control subjects. GTCS consisted of a sequence of phases best described quantitatively by dynamics of the low frequency (LF) wavelet component (2-8Hz). Contrary to the traditional view, the tonic phase started with a gradual increase in muscle activity. A longer clonic phase was associated with a shorter onset of the tonic phase and a higher seizure occurrence. Increase in LF occurred during the onset phase and during the transition from the tonic to the clonic phase, corresponding to the vibratory movements. The clonic phase consisted of EMG discharges of remarkably constant duration (0.2s) separated by silent periods (SP) of exponentially increasing duration - features that could not be reproduced voluntarily. The last SP was longer in seizures with higher EMG peak frequency whereas the energy of the last clonus was higher in seizures with a short clonic phase. We found specific features of muscle activation dynamics during GTCS. Our findings suggest that the same inhibitory mechanisms that contribute to GTCS termination counteract seizure initiation, accounting for the gradual onset. Both active inhibition and mechanisms related to metabolic depletion act synergistically to stop the seizure. Analysis of the ictal EMG dynamics is a valuable tool for monitoring the balance between pro-convulsive and anti-convulsive factors.

Evaluation of novel algorithm embedded in a wearable sEMG device for seizure detection.

We implemented a modified version of a previously published algorithm for detection of generalized tonic-clonic seizures into a prototype wireless surface electromyography (sEMG) recording device. The method was modified to require minimum computational load, and two parameters were trained on prior sEMG data recorded with the device. Along with the normal sEMG recording, the device is able to set an alarm whenever the implemented algorithm detects a seizure. These alarms are annotated in the data file along with the signal. The device was tested at the Epilepsy Monitoring Unit (EMU) at the Danish Epilepsy Center. Five patients were included in the study and two of them had generalized tonic-clonic seizures. All patients were monitored for 2-5 days. A double-blind study was made on the five patients. The overall result showed that the device detected four of seven seizures and had a false detection rate of 0.003/h or one in twelve days.

Automated algorithm for generalized tonic-clonic epileptic seizure onset detection based on sEMG zero-crossing rate.

Patients are not able to call for help during a generalized tonic-clonic epileptic seizure. Our objective was to develop a robust generic algorithm for automatic detection of tonic-clonic seizures, based on surface electromyography (sEMG) signals suitable for a portable device. Twenty-two seizures were analyzed from 11 consecutive patients. Our method is based on a high-pass filtering with a cutoff at 150 Hz, and monitoring a count of zero crossings with a hysteresis of ±50 µV . Based on data from one sEMG electrode (on the deltoid muscle), we achieved a sensitivity of 100% with a mean detection latency of 13.7 s, while the rate of false detection was limited to 1 false alarm per 24 h. The overall performance of the presented generic algorithm is adequate for clinical implementation.

Automatic multi-modal intelligent seizure acquisition (MISA) system for detection of motor seizures from electromyographic data and motion data.

The objective is to develop a non-invasive automatic method for detection of epileptic seizures with motor manifestations. Ten healthy subjects who simulated seizures and one patient participated in the study. Surface electromyography (sEMG) and motion sensor features were extracted as energy measures of reconstructed sub-bands from the discrete wavelet transformation (DWT) and the wavelet packet transformation (WPT). Based on the extracted features all data segments were classified using a support vector machine (SVM) algorithm as simulated seizure or normal activity. A case study of the seizure from the patient showed that the simulated seizures were visually similar to the epileptic one. The multi-modal intelligent seizure acquisition (MISA) system showed high sensitivity, short detection latency and low false detection rate. The results showed superiority of the multi-modal detection system compared to the uni-modal one. The presented system has a promising potential for seizure detection based on multi-modal data.

Seizure onset detection based on one sEMG channel.

We present a new method to detect seizure onsets of tonic-clonic epileptic seizures based on surface electromyography (sEMG) data. The proposed method is generic and based on a single channel making it ideal for a small detection or monitoring device. The sEMG signal is high-pass filtered with a Butterworth filter with a cut-off frequency of 150 Hz. The number of zero-crossings with a hysteresis of ± 50 µV is the only feature extracted. The number of counts in a window of 1 second and the number of windows to make a detection is tested with a leave-one-out method. On 6 patients the method performs with a sensitivity of 100%, a median latency of 7.6 seconds and a median false detection rate of 0.04/h.

Seizure onset detection based on a Uni- or multi-modal intelligent seizure acquisition (UISA/MISA) system.

An automatic Uni- or Multi-modal Intelligent Seizure Acquisition (UISA/MISA) system is highly applicable for onset detection of epileptic seizures based on motion data. The modalities used are surface electromyography (sEMG), acceleration (ACC) and angular velocity (ANG). The new proposed automatic algorithm on motion data is extracting features as "log-sum" measures of discrete wavelet components. Classification into the two groups "seizure" versus "non-seizure" is made based on the support vector machine (SVM) algorithm. The algorithm performs with a sensitivity of 91-100%, a median latency of 1 second and a specificity of 100% on multi-modal data from five healthy subjects simulating seizures. The uni-modal algorithm based on sEMG data from the subjects and patients performs satisfactorily in some cases. As expected, our results clearly show superiority of the multi-modal approach, as compared with the uni-modal one.

Automatic epileptic seizure onset detection using matching pursuit: a case study.

An automatic alarm system for detecting epileptic seizure onsets could be of great assistance to patients and medical staff. A novel approach is proposed using the Matching Pursuit algorithm as a feature extractor combined with the Support Vector Machine (SVM) as a classifier for this purpose. The combination of Matching Pursuit and SVM for automatic seizure detection has never been tested before, making this a pilot study. Data from red different patients with 6 to 49 seizures are used to test our model. Three patients are recorded with scalp electroencephalography (sEEG) and three with intracranial electroencephalography (iEEG). A sensitivity of 78-100% and a detection latency of 5-18s has been achieved, while holding the false detection at 0.16-5.31/h. Our results show the potential of Matching Pursuit as a feature extractor for detection of epileptic seizures.

Automatic seizure detection: going from sEEG to iEEG.

Several different algorithms have been proposed for automatic detection of epileptic seizure based on both scalp and intracranial electroencephalography (sEEG and iEEG). Which modality that renders the best result is hard to assess though. From 16 patients with focal epilepsy, at least 24 hours of ictal and non-ictal iEEG were obtained. Characteristics of the seizures are represented by use of wavelet transformation (WT) features and classified by a support vector machine. When implementing a method used for sEEG on iEEG data, a great improvement in performance was obtained when the high frequency containing lower levels in the WT were included in the analysis. We were able to obtain a sensitivity of 96.4% and a false detection rate (FDR) of 0.20/h. In general, when implementing an automatic seizure detection algorithm made for sEEG on iEEG, great improvement can be obtained if a frequency band widening of the feature extraction is performed. This means that algorithms for sEEG should not be discarded for use on iEEG - they should be properly adjusted as exemplified in this paper.

Multi-modal intelligent seizure acquisition (MISA) system--a new approach towards seizure detection based on full body motion measures.

Many epilepsy patients cannot call for help during a seizure, because they are unconscious or because of the affection of their motor system or speech function. This can lead to injuries, medical complications and at worst death. An alarm system setting off at seizure onset could help to avoid hazards. Today no reliable alarm systems are available. A Multi-modal Intelligent Seizure Acquisition (MISA) system based on full body motion data seems as a good approach towards detection of epileptic seizures. The system is the first to provide a full body description for epilepsy applications. Three test subjects were used for this pilot project. Each subject simulated 15 seizures and in addition performed some predefined normal activities, during a 4-hour monitoring with electromyography (EMG), accelerometer, magnetometer and gyroscope (AMG), electrocardiography (ECG), electroencephalography (EEG) and audio and video recording. The results showed that a non-subject specific MISA system developed on data from the modalities: accelerometer (ACM), gyroscope and EMG is able to detect 98% of the simulated seizures and at the same time mistakes only 4 of the normal movements for seizures. If the system is individualized (subject specific) it is able to detect all simulated seizures with a maximum of 1 false positive. Based on the results from the simulated seizures and normal movements the MISA system seems to be a promising approach to seizure detection.