Jonas Duun-Henriksen

Channel selection for automatic seizure detection

OBJECTIVE To investigate the performance of epileptic seizure detection using only a few of the recorded EEG channels and the ability of software to select these channels compared with a neurophysiologist. METHODS Fifty-nine seizures and 1419 h of interictal EEG are used for training and testing of an automatic channel selection method. The characteristics of the seizures are extracted by the use of a wavelet analysis and classified by a support vector machine. The best channel selection method is based upon maximum variance during the seizure. RESULTS Using only three channels, a seizure detection sensitivity of 96% and a false detection rate of 0.14/h were obtained. This corresponds to the performance obtained when channels are selected through visual inspection by a clinical neurophysiologist, and constitutes a 4% improvement in sensitivity compared to seizure detection using channels recorded directly on the epileptic focus. CONCLUSIONS Based on our dataset, automatic seizure detection can be done using only three EEG channels without loss of performance. These channels should be selected based on maximum variance and not, as often done, using the focal channels. SIGNIFICANCE With this simple automatic channel selection method, we have shown a computational efficient way of making automatic seizure detection.

Differentiating Interictal and Ictal States in Childhood Absence Epilepsy through Permutation Rényi Entropy

Permutation entropy (PE) has been widely exploited to measure the complexity of the electroencephalogram (EEG), especially when complexity is linked to diagnostic information embedded in the EEG. Recently, the authors proposed a spatial-temporal analysis of the EEG recordings of absence epilepsy patients based on PE. The goal here is to improve the ability of PE in discriminating interictal states from ictal states in absence seizure EEG. For this purpose, a parametrical definition of permutation entropy is introduced here in the field of epileptic EEG analysis: the permutation Renyi entropy (PEr). PEr has been extensively tested against PE by tuning the involved parameters (order, delay time and alpha). The achieved results demonstrate that PEr outperforms PE, as there is a statistically-significant, wider gap between the PEr levels during the interictal states and PEr levels observed in the ictal states compared to PE. PEr also outperformed PE as the input to a classifier aimed at discriminating interictal from ictal states.

Automatic detection of childhood absence epilepsy seizures: toward a monitoring device.

Automatic detections of paroxysms in patients with childhood absence epilepsy have been neglected for several years. We acquire reliable detections using only a single-channel brainwave monitor, allowing for unobtrusive monitoring of antiepileptic drug effects. Ultimately we seek to obtain optimal long-term prognoses, balancing antiepileptic effects and side effects. The electroencephalographic appearance of paroxysms in childhood absence epilepsy is fairly homogeneous, making it feasible to develop patient-independent automatic detection. We implemented a state-of-the-art algorithm to investigate the performance of paroxysm detection. Using only a single scalp electroencephalogram channel from 20 patients with a total of 125 paroxysms >2 seconds, 97.2% of paroxysms could be detected with no false detections. This result leads us to recommend further investigations of tiny, one-channel electroencephalogram systems in an ambulatory setting.

Generic single-channel detection of absence seizures

A long-term EEG-monitoring system, which automatically marks seizure events, is useful for diagnosing and treating epilepsy. A generic method utilizing the low inter-and intra-patient variabilities in EEG-characteristics during absence seizures is proposed. This paper investigates if the spike-and-wave behaviour during absence seizures is so distinct that a single-channel implementation is possible. 18 channels of scalp electroencephalography (EEG), from 19 patients suffering from childhood absence epilepsy, are analysed individually. The characteristics of the seizures are captured using the energy content of wavelet transform subbands and classified using a support vector machine. To ease the evaluation of the method, we present a new graphical visualization of the performance based on the topographical distribution on the scalp. The presented seizure detection method shows that the best result is obtained for the derivation F7-FP1. Using this channel a sensitivity of 99.1 %, positive predictive value of 94.8 %, mean detection latency of 3.7 s, and false detection rate value of 0.5/h was obtained. The topographical visualization of the results clearly shows that the frontal, midline, and parietal channels outperform detection based on the channels in the occipital region.

Hypoglycemia-Related Electroencephalogram Changes Assessed by Multiscale Entropy

BACKGROUND Several clinical studies have shown that low blood glucose (BG) levels affect electroencephalogram (EEG) rhythms through the quantification of traditional indicators based on linear spectral analysis. Nonlinear measures used in the last decades to characterize the EEG in several physiopathological conditions have never been assessed in hypoglycemia. The present study investigates if properties of the EEG signal measured by nonlinear entropy-based algorithms are altered in a significant manner when a state of hypoglycemia is entered. SUBJECTS AND METHODS EEG was acquired from 19 patients with type 1 diabetes during a hyperinsulinemic-euglycemic-hypoglycemic clamp experiment. In parallel, BG was frequently monitored by the standard YSI glucose and lactate analyzer and used to identify two 1-h intervals corresponding to euglycemia and hypoglycemia, respectively. In each subject, the P3-C3 EEG derivation in the two glycemic intervals was assessed using the multiscale entropy (MSE) approach, obtaining measures of sample entropy (SampEn) at various temporal scales. The comparison of how signal irregularity measured by SampEn varies as the temporal scale increases in the two glycemic states provides information on how EEG complexity is affected by hypoglycemia. RESULTS For both glycemic states, the MSE analysis showed that SampEn increases at small time scales and then monotonically decreases as the time scale becomes larger. Comparing the two conditions, SampEn was higher in hypoglycemia only at medium time scales. CONCLUSIONS A decrease in the complexity of EEG occurs when a state of hypoglycemia is entered, because of a degradation of the EEG long-range temporal correlations. Thanks to its ability to assess nonlinear dynamics of the EEG signal, the MSE approach seems to be a useful tool to complement information brought by standard linear indicators and provide new insights on how hypoglycemia affects brain functioning.

EEG Signal Quality of a Subcutaneous Recording System Compared to Standard Surface Electrodes

Purpose. We provide a comprehensive verification of a new subcutaneous EEG recording device which promises robust and unobtrusive measurements over ultra-long time periods. The approach is evaluated against a state-of-the-art surface EEG electrode technology. Materials and Methods. An electrode powered by an inductive link was subcutaneously implanted on five subjects. Surface electrodes were placed at sites corresponding to the subcutaneous electrodes, and the EEG signals were evaluated with both quantitative (power spectral density and coherence analysis) and qualitative (blinded subjective scoring by neurophysiologists) analysis. Results. The power spectral density and coherence analysis were very similar during measurements of resting EEG. The scoring by neurophysiologists showed a higher EEG quality for the implanted system for different subject states (eyes open and eyes closed). This was most likely due to higher amplitude of the subcutaneous signals. During periods with artifacts, such as chewing, blinking, and eye movement, the two systems performed equally well. Conclusions. Subcutaneous measurements of EEG with the test device showed high quality as measured by both quantitative and more subjective qualitative methods. The signal might be superior to surface EEG in some aspects and provides a method of ultra-long term EEG recording in situations where this is required and where a small number of EEG electrodes are sufficient.

Subdural to subgaleal EEG signal transmission: The role of distance, leakage and insulating affectors

OBJECTIVE To estimate the area of cortex affecting the extracranial EEG signal. METHODS The coherence between intra- and extracranial EEG channels were evaluated on at least 10 min of spontaneous, awake data from seven patients admitted for epilepsy surgery work up. RESULTS Cortical electrodes showed significant extracranial coherent signals in an area of approximately 150 cm(2) although the field of vision was probably only 31 cm(2) based on spatial averaging of intracranial channels taking into account the influence of the craniotomy and the silastic membrane of intracranial grids. Selecting the best cortical channels, it was possible to increase the coherence values compared to the single intracranial channel with highest coherence. The coherence seemed to increase linearly with an accumulation area up to 31 cm(2), where 50% of the maximal coherence was obtained accumulating from only 2 cm(2) (corresponding to one channel), and 75% when accumulating from 16 cm(2). CONCLUSION The skull is an all frequency spatial averager but dominantly high frequency signal attenuator. SIGNIFICANCE An empirical assessment of the actual area of cerebral sources generating the extracranial EEG provides better opportunities for clinical electroencephalographers to determine the location of origin of particular patterns in the EEG.

Wavelet coherence-based clustering of EEG signals to estimate the brain connectivity in absence epileptic patients

In this paper, the need of novel methods to extract diagnostic information from the Electroencephalographic (EEG) recordings of epileptic patients was addressed. A novel method, based on Wavelet Coherence (WC) between EEG signals and Hierarchical Clustering (HC), was proposed to estimate the EEG network connectivity density in Childhood Absence Epilepsy (CAE) patients. The EEG recordings of four patients affected by CAE were partitioned into non overlapping windows and WC was estimated window by window. The behaviour of WC was analysed over the time, for every couple of EEG electrodes. The ictal states (seizures) resulted associated to increased WC levels, thus reflecting an increased synchronization between electrodes during the seizure. A WC-based dissimilarity index was then defined and HC was fed with the dissimilarity indices between every pair of electrodes with the aim of finding possible correlations between changes in electrode clustering and changes in the brain state. For every window under analysis, a dendrogram was constructed, the corresponding set of electrode clusters was determined and the subsequent network density values were calculated. Seizures resulted typically associated to increased network density, reflecting an increased connectivity during the ictal states.

Quantifying the Complexity of Epileptic EEG

In this paper, the issue of automatic epileptic seizure detection is addressed, emphasizing how the huge amount of Electroencephalographic (EEG) data from epileptic patients can slow down the diagnostic procedure and cause mistakes. The EEG of an epileptic patient can last from minutes to many hours and the goal here is to automatically detect the seizures that occurr during the EEG recording. In other words, the goal is to automatically discriminate between the interictal and ictal states of the brain so that the neurologist can immediately focus on the ictal states with no need of detecting such events manually. In particular, the attention is focused on absence seizures. The goal is to develop a system that is able to extract meaningful features from the EEG and to learn how to classify the brain states accordingly. The complexity of the EEG is considered a key feature when dealing with an epileptic brain and two measures of complexity are here estimated and compared in the task of interictal-ictal states discrimination: Approximate Entropy (ApEn) and Permutation Entropy (PE). A Learning Vector Quantization network is then fed with ApEn and PE and trained. The ApEn+LVQ learning system provided a better sensitivity compared to the PE+LVQ one, nevertheless, it showed a smaller specificity.

Detection of Paroxysms in Long-Term, Single-Channel EEG-Monitoring of Patients with Typical Absence Seizures

Absence seizures are associated with generalized 2.5–5 Hz spike-wave discharges in the electroencephalogram (EEG). Rarely are patients, parents, or physicians aware of the duration or incidence of seizures. Six patients were monitored with a portable EEG-device over four times 24 h to evaluate how easily outpatients are monitored and how well an automatic seizure detection algorithm can identify the absences. Based on patient-specific modeling, we achieved a sensitivity of 98.4% with only 0.23 false detections per hour. This yields a clinically satisfying performance with a positive predictive value of 87.1%. Portable EEG-recorders identifying paroxystic events in epilepsy outpatients are a promising tool for patients and physicians dealing with absence epilepsy. Albeit the small size of the EEG-device, some children still complained about the obtrusive nature of the device. We aim at developing less obtrusive though still very efficient devices, e.g., hidden in the ear canal or below the skin.