A multistage system for the automated detection of epileptic seizures in neonatal electroencephalography

Quantifying Sharpness and Nonlinearity in Neonatal Seizure Dynamics

The integration of multiple electrophysiological biomarkers is crucial for monitoring neonatal seizure dynamics. The present study aimed to characterize the temporal dynamics of neonatal seizures by analyzing intrinsic waveforms of epileptic electroencephalogram (EEG) signals. We proposed a complementary set of methods considering envelope power, focal sharpness changes, and nonlinear patterns of EEG signals of 79 neonates with seizures. Features derived from EEG signals were used as input to the machine learning classifier. All three characteristics were significantly elevated during seizure events, as agreed upon by all viewers (P < 0.0001). Envelope power was elevated in the entire seizure period, and the degree of nonlinearity rose at the termination of a seizure event. Epileptic sharpness effectively characterizes an entire seizure event, complementing the role of envelope power in identifying its onset. However, the degree of nonlinearity showed superior discriminability for the termination of a seizure event. The proposed computational methods for intrinsic sharp or nonlinear EEG patterns evolving during neonatal seizure could share some features with envelope power. Current findings may be helpful in developing strategies to improve neonatal seizure monitoring.

XAI4EEG: spectral and spatio-temporal explanation of deep learning-based seizure detection in EEG time series

In clinical practice, algorithmic predictions may seriously jeopardise patients’ health and thus are required to be validated by medical experts before a final clinical decision is met. Towards that aim, there is need to incorporate explainable artificial intelligence techniques into medical research. In the specific field of epileptic seizure detection there are several machine learning algorithms but less methods on explaining them in an interpretable way. Therefore, we introduce XAI4EEG: an application-aware approach for an explainable and hybrid deep learning-based detection of seizures in multivariate EEG time series. In XAI4EEG, we combine deep learning models and domain knowledge on seizure detection, namely (a) frequency bands, (b) location of EEG leads and (c) temporal characteristics. XAI4EEG encompasses EEG data preparation, two deep learning models and our proposed explanation module visualizing feature contributions that are obtained by two SHAP explainers, each explaining the predictions of one of the two models. The resulting visual explanations provide an intuitive identification of decision-relevant regions in the spectral, spatial and temporal EEG dimensions. To evaluate XAI4EEG, we conducted a user study, where users were asked to assess the outputs of XAI4EEG, while working under time constraints, in order to emulate the fact that clinical diagnosis is done - more often than not - under time pressure. We found that the visualizations of our explanation module (1) lead to a substantially lower time for validating the predictions and (2) leverage an increase in interpretability, trust and confidence compared to selected SHAP feature contribution plots.

Neonatal EEG seizure detection using a new signal structural complexity measure based on matching pursuit decomposition with nonstationary dictionary

BACKGROUND AND OBJECTIVE

In newborns, it is often difficult to accurately differentiate between seizure and non-seizure based solely on clinical manifestations. This highlights the importance of electroencephalogram (EEG) in the recognition and management of neonatal seizures. This paper proposes an effective algorithm for the detection of neonatal seizure using multichannel EEG.

METHODS

Neonatal EEG changes morphology as it alternates between seizure and non-seizure states. A new signal complexity measure based on matching pursuit (MP) decomposition is proposed and used to detect transitions between these two states. The new measure, referred to as weighted structural complexity (WSC), was used for the detection of seizures in 30 newborn EEG records. Multiple IIR filters and an MP-based filter were designed and used to remove artifacts from the EEG data. Geometrical correlation between the EEG data channels was applied to reduce the number of false detections caused by remnant artifacts. The seizure detector's performance was assessed using several epoch-based (e.g., accuracy) and event-based (GDR = good detection rate and FD/h = false detections per hour) metrics.

RESULTS

Compared to the neurologist marking, the proposed detector was able to detect EEG seizures with 94% accuracy, 90.9% GDR, and 0.14 FD/h (95% CI: [0.06, 0.34]).

CONCLUSIONS

The high performance of the MP-based detector may have significant implications for the accurate diagnosis of neonatal seizures and the appropriate use of anticonvulsants and ongoing clinical assessment and care of the newborn.

Optical Flow Estimation Improves Automated Seizure Detection in Neonatal EEG

PURPOSE

Existing automated seizure detection algorithms report sensitivities between 43% and 77% and specificities between 56% and 90%. The algorithms suffer from false alarms when applied to neonatal EEG because of the high degree of nurse handling and rhythmic patting used to soothe neonates. Computer vision technology that quantifies movement in real time could distinguish artifactual motion and improve automated neonatal seizure detection algorithms.

METHODS

The authors used video EEG recordings from 43 neonates undergoing monitoring for seizures as part of the NEOLEV2 clinical trial. The Persyst neonatal automated seizure detection algorithm ran in real time during study EEG acquisitions. Computer vision algorithms were applied to extract detailed accounts of artifactual movement of the neonate or people near the neonate though dense optical flow estimation.

RESULTS

Using the methods mentioned above, 197 periods of patting activity were identified and quantified, of which 45 generated false-positive automated seizure detection events. A binary patting detection algorithm was trained with a subset of 470 event videos. This supervised detection algorithm was applied to a testing subset of 187 event videos with 8 false-positive events, which resulted in a 24% reduction in false-positive automated seizure detections and a 50% reduction in false-positive events caused by neonatal care patting, while maintaining 11 of 12 true-positive seizure detection events.

CONCLUSIONS

This work presents a novel approach to improving automated seizure detection algorithms used during neonatal video EEG monitoring. This artifact detection mechanism can improve the ability of a seizure detector algorithm to distinguish between artifact and true seizure activity.

Epileptic-Net: An Improved Epileptic Seizure Detection System Using Dense Convolutional Block with Attention Network from EEG

Epilepsy is a complex neurological condition that affects a large number of people worldwide. Electroencephalography (EEG) measures the electrical activity of the brain and is widely used in epilepsy diagnosis, but it usually requires manual inspection, which can be hours long, by a neurologist. Several automatic systems have been proposed to detect epilepsy but still have some unsolved issues. In this study, we proposed a dynamic method using a deep learning model (Epileptic-Net) to detect an epileptic seizure. The proposed method is largely heterogeneous and comprised of the dense convolutional blocks (DCB), feature attention modules (FAM), residual blocks (RB), and hypercolumn technique (HT). Firstly, DCB is used to get the discriminative features from the EEG samples. Then, FAM extracts the essential features from the samples. After that, RB learns more vital parts as it entirely uses information in the convolutional layer. Finally, HT retains the efficient local features extracted from the layers situated at the different levels of the model. Its performance has been evaluated on the University of Bonn EEG dataset, divided into five distinct classes. The proposed Epileptic-Net achieves the average accuracy of 99.95% in the two-class classification, 99.98% in the three-class classification, 99.96% in the four-class classification, and 99.96% in classifying the complicated five-class problem. Thus the proposed approach shows more competitive results than the existing model to detect epileptic seizures. We also hope that this method can support experts in achieving objective and reliable results, lowering the misdiagnosis rate, and assisting in decision-making.

Convolutional neural networks ensemble model for neonatal seizure detection

BACKGROUND

Neonatal seizures are a common occurrence in clinical settings, requiring immediate attention and detection. Previous studies have proposed using manual feature extraction coupled with machine learning, or deep learning to classify between seizure and non-seizure states.

NEW METHOD

In this paper a deep learning based approach is used for neonatal seizure classification using electroencephalogram (EEG) signals. The architecture detects seizure activity in raw EEG signals as opposed to common state-of-art, where manual feature extraction with machine learning algorithms is used. The architecture is a two-dimensional (2D) convolutional neural network (CNN) to classify between seizure/non-seizure states.

RESULTS

The dataset used for this study is annotated by three experts and as such three separate models are trained on individual annotations, resulting in average accuracies (ACC) of 95.6 %, 94.8 % and 90.1 % respectively, and average area under the receiver operating characteristic curve (AUC) of 99.2 %, 98.4 % and 96.7 % respectively. The testing was done using 10-cross fold validation, so that the performance can be an accurate representation of the architectures classification capability in a clinical setting. After training/testing of the three individual models, a final ensemble model is made consisting of the three models. The ensemble model gives an average ACC and AUC of 96.3 % and 99.3 % respectively.

COMPARISON WITH EXISTING METHODS

This study outperforms previous studies, with increased ACC and AUC results coupled with use of small time windows (1 s) used for evaluation.

CONCLUSION

The proposed approach is promising for detecting seizure activity in unseen neonate data in a clinical setting.

A machine-learning algorithm for neonatal seizure recognition: a multicentre, randomised, controlled trial

BACKGROUND

Despite the availability of continuous conventional electroencephalography (cEEG), accurate diagnosis of neonatal seizures is challenging in clinical practice. Algorithms for decision support in the recognition of neonatal seizures could improve detection. We aimed to assess the diagnostic accuracy of an automated seizure detection algorithm called Algorithm for Neonatal Seizure Recognition (ANSeR).

METHODS

This multicentre, randomised, two-arm, parallel, controlled trial was done in eight neonatal centres across Ireland, the Netherlands, Sweden, and the UK. Neonates with a corrected gestational age between 36 and 44 weeks with, or at significant risk of, seizures requiring EEG monitoring, received cEEG plus ANSeR linked to the EEG monitor displaying a seizure probability trend in real time (algorithm group) or cEEG monitoring alone (non-algorithm group). The primary outcome was diagnostic accuracy (sensitivity, specificity, and false detection rate) of health-care professionals to identify neonates with electrographic seizures and seizure hours with and without the support of the ANSeR algorithm. Neonates with data on the outcome of interest were included in the analysis. This study is registered with ClinicalTrials.gov, NCT02431780.

FINDINGS

Between Feb 13, 2015, and Feb 7, 2017, 132 neonates were randomly assigned to the algorithm group and 132 to the non-algorithm group. Six neonates were excluded (four from the algorithm group and two from the non-algorithm group). Electrographic seizures were present in 32 (25·0%) of 128 neonates in the algorithm group and 38 (29·2%) of 130 neonates in the non-algorithm group. For recognition of neonates with electrographic seizures, sensitivity was 81·3% (95% CI 66·7-93·3) in the algorithm group and 89·5% (78·4-97·5) in the non-algorithm group; specificity was 84·4% (95% CI 76·9-91·0) in the algorithm group and 89·1% (82·5-94·7) in the non-algorithm group; and the false detection rate was 36·6% (95% CI 22·7-52·1) in the algorithm group and 22·7% (11·6-35·9) in the non-algorithm group. We identified 659 h in which seizures occurred (seizure hours): 268 h in the algorithm versus 391 h in the non-algorithm group. The percentage of seizure hours correctly identified was higher in the algorithm group than in the non-algorithm group (177 [66·0%; 95% CI 53·8-77·3] of 268 h vs 177 [45·3%; 34·5-58·3] of 391 h; difference 20·8% [3·6-37·1]). No significant differences were seen in the percentage of neonates with seizures given at least one inappropriate antiseizure medication (37·5% [95% CI 25·0 to 56·3] vs 31·6% [21·1 to 47·4]; difference 5·9% [-14·0 to 26·3]).

INTERPRETATION

ANSeR, a machine-learning algorithm, is safe and able to accurately detect neonatal seizures. Although the algorithm did not enhance identification of individual neonates with seizures beyond conventional EEG, recognition of seizure hours was improved with use of ANSeR. The benefit might be greater in less experienced centres, but further study is required.

FUNDING

Wellcome Trust, Science Foundation Ireland, and Nihon Kohden.

Convolutional neural network with autoencoder-assisted multiclass labelling for seizure detection based on scalp electroencephalography

OBJECTIVE

In long-term video-monitoring, automatic seizure detection holds great promise as a means to reduce the workload of the epileptologist. A convolutional neural network (CNN) designed to process images of EEG plots demonstrated high performance for seizure detection, but still has room for reducing the false-positive alarm rate.

METHODS

We combined a CNN that processed images of EEG plots with patient-specific autoencoders (AE) of EEG signals to reduce the false alarms during seizure detection. The AE automatically logged abnormalities, i.e., both seizures and artifacts. Based on seizure logs compiled by expert epileptologists and errors made by AE, we constructed a CNN with 3 output classes: seizure, non-seizure-but-abnormal, and non-seizure. The accumulative measure of number of consecutive seizure labels was used to issue a seizure alarm.

RESULTS

The second-by-second classification performance of AE-CNN was comparable to that of the original CNN. False-positive seizure labels in AE-CNN were more likely interleaved with "non-seizure-but-abnormal" labels than with true-positive seizure labels. Consequently, "non-seizure-but-abnormal" labels interrupted runs of false-positive seizure labels before triggering an alarm. The median false alarm rate with the AE-CNN was reduced to 0.034 h^-1, which was one-fifth of that of the original CNN (0.17 h^-1).

CONCLUSIONS

A label of "non-seizure-but-abnormal" offers practical benefits for seizure detection. The modification of a CNN with an AE is worth considering because AEs can automatically assign "non-seizure-but-abnormal" labels in an unsupervised manner with no additional demands on the time of the epileptologist.

Applications of advanced signal processing and machine learning in the neonatal hypoxic-ischemic electroencephalogram

Perinatal hypoxic-ischemic-encephalopathy significantly contributes to neonatal death and life-long disability such as cerebral palsy. Advances in signal processing and machine learning have provided the research community with an opportunity to develop automated real-time identification techniques to detect the signs of hypoxic-ischemic-encephalopathy in larger electroencephalography/amplitude-integrated electroencephalography data sets more easily. This review details the recent achievements, performed by a number of prominent research groups across the world, in the automatic identification and classification of hypoxic-ischemic epileptiform neonatal seizures using advanced signal processing and machine learning techniques. This review also addresses the clinical challenges that current automated techniques face in order to be fully utilized by clinicians, and highlights the importance of upgrading the current clinical bedside sampling frequencies to higher sampling rates in order to provide better hypoxic-ischemic biomarker detection frameworks. Additionally, the article highlights that current clinical automated epileptiform detection strategies for human neonates have been only concerned with seizure detection after the therapeutic latent phase of injury. Whereas recent animal studies have demonstrated that the latent phase of opportunity is critically important for early diagnosis of hypoxic-ischemic-encephalopathy electroencephalography biomarkers and although difficult, detection strategies could utilize biomarkers in the latent phase to also predict the onset of future seizures.

Time-Varying EEG Correlations Improve Automated Neonatal Seizure Detection

The aim of this study was to develop methods for detecting the nonstationary periodic characteristics of neonatal electroencephalographic (EEG) seizures by adapting estimates of the correlation both in the time (spike correlation; SC) and time–frequency domain (time–frequency correlation; TFC). These measures were incorporated into a seizure detection algorithm (SDA) based on a support vector machine to detect periods of seizure and nonseizure. The performance of these nonstationary correlation measures was evaluated using EEG recordings from 79 term neonates annotated by three human experts. The proposed measures were highly discriminative for seizure detection (median [Formula: see text]: 0.933 IQR: 0.821–0.975, median [Formula: see text]: 0.883 IQR: 0.707–0.931). The resultant SDA applied to multi-channel recordings had a median AUC of 0.988 (IQR: 0.931–0.998) when compared to consensus annotations, outperformed two state-of-the-art SDAs [Formula: see text] and was noninferior to the human expert for 73/79 of neonates.

Neonatal Seizure Detection Using Deep Convolutional Neural Networks

Identifying a core set of features is one of the most important steps in the development of an automated seizure detector. In most of the published studies describing features and seizure classifiers, the features were hand-engineered, which may not be optimal. The main goal of the present paper is using deep convolutional neural networks (CNNs) and random forest to automatically optimize feature selection and classification. The input of the proposed classifier is raw multi-channel EEG and the output is the class label: seizure/nonseizure. By training this network, the required features are optimized, while fitting a nonlinear classifier on the features. After training the network with EEG recordings of 26 neonates, five end layers performing the classification were replaced with a random forest classifier in order to improve the performance. This resulted in a false alarm rate of 0.9 per hour and seizure detection rate of 77% using a test set of EEG recordings of 22 neonates that also included dubious seizures. The newly proposed CNN classifier outperformed three data-driven feature-based approaches and performed similar to a previously developed heuristic method.

A dataset of neonatal EEG recordings with seizure annotations

Neonatal seizures are a common emergency in the neonatal intensive care unit (NICU). There are many questions yet to be answered regarding the temporal/spatial characteristics of seizures from different pathologies, response to medication, effects on neurodevelopment and optimal detection. The dataset presented in this descriptor contains EEG recordings from human neonates, the visual interpretation of the EEG by the human experts, supporting clinical data and codes to assist access. Multi-channel EEG was recorded from 79 term neonates admitted to the NICU at the Helsinki University Hospital. The median recording duration was 74 min (IQR: 64 to 96 min). The presence of seizures in the EEGs was annotated independently by three experts. An average of 460 seizures were annotated per expert in the dataset; 39 neonates had seizures and 22 were seizure free, by consensus. The dataset can be used as a reference set of neonatal seizures, in studies of inter-observer agreement and for the development of automated methods of seizure detection and other EEG analyses.

Diagnosis of seizures and encephalopathy using conventional EEG and amplitude integrated EEG

Seizures are more common in the neonatal period than at any other time of life, partly due to the relative hyperexcitability of the neonatal brain. Brain monitoring of sick neonates in the NICU using either conventional electroencephalography or amplitude integrated EEG is essential to accurately detect seizures. Treatment of seizures is important, as evidence increasingly indicates that seizures damage the brain in addition to that caused by the underlying etiology. Prompt treatment has been shown to reduce seizure burden with the potential to ameliorate seizure-mediated damage. Neonatal encephalopathy most commonly caused by a hypoxia-ischemia results in an alteration of mental status and problems such as seizures, hypotonia, apnea, and feeding difficulties. Confirmation of encephalopathy with EEG monitoring can act as an important adjunct to other investigations and the clinical examination, particularly when considering treatment strategies such as therapeutic hypothermia. Brain monitoring also provides useful early prognostic indicators to clinicians. Recent use of machine learning in algorithms to continuously monitor the neonatal EEG, detect seizures, and grade encephalopathy offers the exciting prospect of real-time decision support in the NICU in the very near future.

Toward a Personalized Real-Time Diagnosis in Neonatal Seizure Detection

The problem of creating a personalized seizure detection algorithm for newborns is tackled in this paper. A probabilistic framework for semi-supervised adaptation of a generic patient-independent neonatal seizure detector is proposed. A system that is based on a combination of patient-adaptive (generative) and patient-independent (discriminative) classifiers is designed and evaluated on a large database of unedited continuous multichannel neonatal EEG recordings of over 800 h in duration. It is shown that an improvement in the detection of neonatal seizures over the course of long EEG recordings is achievable with on-the-fly incorporation of patient-specific EEG characteristics. In the clinical setting, the employment of the developed system will maintain a seizure detection rate at 70% while halving the number of false detections per hour, from 0.4 to 0.2 FD/h. This is the first study to propose the use of online adaptation without clinical labels, to build a personalized diagnostic system for the detection of neonatal seizures.

EEG-Based Brain-Computer Interface for Decoding Motor Imagery Tasks within the Same Hand Using Choi-Williams Time-Frequency Distribution

This paper presents an EEG-based brain-computer interface system for classifying eleven motor imagery (MI) tasks within the same hand. The proposed system utilizes the Choi-Williams time-frequency distribution (CWD) to construct a time-frequency representation (TFR) of the EEG signals. The constructed TFR is used to extract five categories of time-frequency features (TFFs). The TFFs are processed using a hierarchical classification model to identify the MI task encapsulated within the EEG signals. To evaluate the performance of the proposed approach, EEG data were recorded for eighteen intact subjects and four amputated subjects while imagining to perform each of the eleven hand MI tasks. Two performance evaluation analyses, namely channel- and TFF-based analyses, are conducted to identify the best subset of EEG channels and the TFFs category, respectively, that enable the highest classification accuracy between the MI tasks. In each evaluation analysis, the hierarchical classification model is trained using two training procedures, namely subject-dependent and subject-independent procedures. These two training procedures quantify the capability of the proposed approach to capture both intra- and inter-personal variations in the EEG signals for different MI tasks within the same hand. The results demonstrate the efficacy of the approach for classifying the MI tasks within the same hand. In particular, the classification accuracies obtained for the intact and amputated subjects are as high as 88.8% and 90.2%, respectively, for the subject-dependent training procedure, and 80.8% and 87.8%, respectively, for the subject-independent training procedure. These results suggest the feasibility of applying the proposed approach to control dexterous prosthetic hands, which can be of great benefit for individuals suffering from hand amputations.

Phenobarbital reduces EEG amplitude and propagation of neonatal seizures but does not alter performance of automated seizure detection

•

Phenobarbital reduces both amplitude and propagation of neonatal seizures.

•

These changes may help to explain electroclinical uncoupling.

•

The performance of our seizure detection algorithm was unaffected.

Objective

Phenobarbital increases electroclinical uncoupling and our preliminary observations suggest it may also affect electrographic seizure morphology. This may alter the performance of a novel seizure detection algorithm (SDA) developed by our group.

The objectives of this study were to compare the morphology of seizures before and after phenobarbital administration in neonates and to determine the effect of any changes on automated seizure detection rates.

Methods

The EEGs of 18 term neonates with seizures both pre- and post-phenobarbital (524 seizures) administration were studied. Ten features of seizures were manually quantified and summary measures for each neonate were statistically compared between pre- and post-phenobarbital seizures. SDA seizure detection rates were also compared.

Results

Post-phenobarbital seizures showed significantly lower amplitude (p < 0.001) and involved fewer EEG channels at the peak of seizure (p < 0.05). No other features or SDA detection rates showed a statistical difference.

Conclusion

These findings show that phenobarbital reduces both the amplitude and propagation of seizures which may help to explain electroclinical uncoupling of seizures. The seizure detection rate of the algorithm was unaffected by these changes.

Significance

The results suggest that users should not need to adjust the SDA sensitivity threshold after phenobarbital administration.

Recording of Neonatal Seizures in Birth Certificates, Maternal Interviews, and Hospital Discharge Abstracts in a Cerebral Palsy Case-Control Study in Michigan

We evaluated the recording of neonatal seizures in birth certificates, hospital discharge abstracts, and maternal interviews in 372 children, 198 of them with cerebral palsy, born in Michigan hospitals from 1993 to 2010. In birth certificates, we examined checkbox items "seizures" or "seizure or serious neurologic dysfunction"; in hospital discharge abstracts ICD-9-CM codes 779.0, 345.X, and 780.3; and in maternal interviews a history of seizures or convulsions on day 1 of life recalled 2-16 years later. In 27 neonates, 38 neonatal seizures were recorded in 1 or more sources, 17 in discharge abstracts, 20 in maternal interviews, but just 1 on a birth certificate. The kappa coefficient (κ) between interviews and discharge abstracts was moderate (κ = 0.55), and substantial (κ = 0.63) if mothers noted use of antiepileptics. Agreement was higher (κ = 0.71 vs κ = 0.29) in term births than in preterm births. Birth certificates significantly underreported neonatal seizures.

In-depth performance analysis of an EEG based neonatal seizure detection algorithm

•

A novel method for in-depth analysis of neonatal seizure detection algorithms is proposed.

•

The analysis estimated how seizure features are exploited by automated detectors.

•

This method led to significant improvement of the ANSeR algorithm.

Objective

To describe a novel neurophysiology based performance analysis of automated seizure detection algorithms for neonatal EEG to characterize features of detected and non-detected seizures and causes of false detections to identify areas for algorithmic improvement.

Methods

EEGs of 20 term neonates were recorded (10 seizure, 10 non-seizure). Seizures were annotated by an expert and characterized using a novel set of 10 criteria.

ANSeR seizure detection algorithm (SDA) seizure annotations were compared to the expert to derive detected and non-detected seizures at three SDA sensitivity thresholds. Differences in seizure characteristics between groups were compared using univariate and multivariate analysis. False detections were characterized.

Results

The expert detected 421 seizures. The SDA at thresholds 0.4, 0.5, 0.6 detected 60%, 54% and 45% of seizures. At all thresholds, multivariate analyses demonstrated that the odds of detecting seizure increased with 4 criteria: seizure amplitude, duration, rhythmicity and number of EEG channels involved at seizure peak. Major causes of false detections included respiration and sweat artefacts or a highly rhythmic background, often during intermediate sleep.

Conclusion

This rigorous analysis allows estimation of how key seizure features are exploited by SDAs.

Significance

This study resulted in a beta version of ANSeR with significantly improved performance.

Interobserver agreement for neonatal seizure detection using multichannel EEG

Objective

To determine the interobserver agreement (IOA) of neonatal seizure detection using the gold standard of conventional, multichannel EEG.

Methods

A cohort of full‐term neonates at risk of acute encephalopathy was included in this prospective study. The EEG recordings of these neonates were independently reviewed for seizures by three international experts. The IOA was estimated using statistical measures including Fleiss' kappa and percentage agreement assessed over seizure events (event basis) and seizure duration (temporal basis).

Results

A total of 4066 h of EEG recordings from 70 neonates were reviewed with an average of 2555 seizures detected. The IOA was high with temporal assessment resulting in a kappa of 0.827 (95% CI: 0.769–0.865; n = 70). The median agreement was 83.0% (interquartile range [IQR]: 76.6–89.5%; n = 33) for seizure and 99.7% (IQR: 98.9–99.8%; n = 70) for nonseizure EEG. Analysis of events showed a median agreement of 83.0% (IQR: 72.9–86.6%; n = 33) for seizures with 0.018 disagreements per hour (IQR: 0.000–0.090 per hour; n = 70). Observers were more likely to disagree when a seizure was less than 30 sec. Overall, 33 neonates were diagnosed with seizures and 28 neonates were not, by all three observers. Of the remaining nine neonates with contradictory EEG detections, seven presented with low total seizure burden.

Interpretation

The IOA is high among experts for the detection of neonatal seizures using conventional, multichannel EEG. Agreement is reduced when seizures are rare or have short duration. These findings support EEG‐based decision making in the neonatal intensive care unit, inform EEG interpretation guidelines, and provide benchmarks for seizure detection algorithms.

A neural network based infant monitoring system to facilitate diagnosis of epileptic seizures

In this paper, we propose an infant monitoring system that automatically detects epileptic seizures in domestic and hospital environments. The proposed system measures the movements and electroencephalogram (EEG) signals of an infant using a video camera and an electroencephalograph. Seizure features are then extracted from the video images and EEG signals, and the evaluation indices based on medical knowledge are calculated from the features. The system employs a probabilistic neural network for the automatic detection of seizures, thereby allowing the choice/combination of evaluation indices appropriate for the environment via network training. We tested the system in simulated domestic and hospital environments. The validity of the proposed system was reinforced by the results of comparisons with clinical diagnoses.

Validation of an automated seizure detection algorithm for term neonates

OBJECTIVE

The objective of this study was to validate the performance of a seizure detection algorithm (SDA) developed by our group, on previously unseen, prolonged, unedited EEG recordings from 70 babies from 2 centres.

METHODS

EEGs of 70 babies (35 seizure, 35 non-seizure) were annotated for seizures by experts as the gold standard. The SDA was tested on the EEGs at a range of sensitivity settings. Annotations from the expert and SDA were compared using event and epoch based metrics. The effect of seizure duration on SDA performance was also analysed.

RESULTS

Between sensitivity settings of 0.5 and 0.3, the algorithm achieved seizure detection rates of 52.6-75.0%, with false detection (FD) rates of 0.04-0.36FD/h for event based analysis, which was deemed to be acceptable in a clinical environment. Time based comparison of expert and SDA annotations using Cohen's Kappa Index revealed a best performing SDA threshold of 0.4 (Kappa 0.630). The SDA showed improved detection performance with longer seizures.

CONCLUSION

The SDA achieved promising performance and warrants further testing in a live clinical evaluation.

SIGNIFICANCE

The SDA has the potential to improve seizure detection and provide a robust tool for comparing treatment regimens.

Epileptic Seizure Detection Based on Partial Directed Coherence Analysis

Long-term video EEG epilepsy monitoring can help doctors diagnose and cure epilepsy. The workload of doctors to read the EEG signals of epilepsy patients can be effectively reduced by automatic seizure detection. The application of partial directed coherence (PDC) analysis as mechanism for feature extraction in the scalp EEG recordings for seizure detection could reflect the physiological changes of brain activity before and after seizure onsets. In this study, a new approach on the basis of PDC was proposed to detect the seizure intervals of epilepsy patients. First of all, the multivariate autoregressive model was established for a moving window and the direction and intensity of information flow based on PDC analysis was calculated. Then, the outflow information related to certain EEG channel could be obtained by summing up the intensity of information flow propagated to other EEG channels in order to reduce the feature dimensionality. At last, according to the pathological features of epileptic seizures, the outflow information was regarded as the input vectors to a support vector machine classifier for discriminating interictal periods and ictal periods of EEG signals. The proposed method had achieved a good performance with the correct rate of 98.3%, the selectivity rate of 67.88%, the sensitivity rate of 91.44%, the specificity rate of 99.34%, and the average detection rate of 95.39%, which demonstrated that this method was suitable for detecting the seizure intervals of epilepsy patients. By comparing with other existing techniques, the proposed method based on PDC analysis achieved significant improvement in terms of seizure detection.

Neonatal seizure detection using atomic decomposition with a novel dictionary

Atomic decomposition (AD) can be used to efficiently decompose an arbitrary signal. In this paper, we present a method to detect neonatal electroencephalogram (EEG) seizure based on AD via orthogonal matching pursuit using a novel, application-specific, dictionary. The dictionary consists of pseudoperiodic Duffing oscillator atoms which are designed to be coherent with the seizure epochs. The relative structural complexity (a measure of the rate of convergence of AD) is used as the sole feature for seizure detection. The proposed feature was tested on a large clinical dataset of 826 h of EEG data from 18 full-term newborns with 1389 seizures. The seizure detection system using the proposed dictionary was able to achieve a median receiver operator characteristic area of 0.91 (IQR 0.87-0.95) across 18 neonates.

Clinical implementation of a neonatal seizure detection algorithm

Technologies for automated detection of neonatal seizures are gradually moving towards cot-side implementation. The aim of this paper is to present different ways to visualize the output of a neonatal seizure detection system and analyse their influence on performance in a clinical environment. Three different ways to visualize the detector output are considered: a binary output, a probabilistic trace, and a spatio-temporal colormap of seizure observability. As an alternative to visual aids, audified neonatal EEG is also considered. Additionally, a survey on the usefulness and accuracy of the presented methods has been performed among clinical personnel. The main advantages and disadvantages of the presented methods are discussed. The connection between information visualization and different methods to compute conventional metrics is established. The results of the visualization methods along with the system validation results indicate that the developed neonatal seizure detector with its current level of performance would unambiguously be of benefit to clinicians as a decision support system. The results of the survey suggest that a suitable way to visualize the output of neonatal seizure detection systems in a clinical environment is a combination of a binary output and a probabilistic trace. The main healthcare benefits of the tool are outlined. The decision support system with the chosen visualization interface is currently undergoing pre-market European multi-centre clinical investigation to support its regulatory approval and clinical adoption.

Diagnosing neonatal seizures and status epilepticus

Continuous electroencephalographic (CEEG) monitoring is often applied in the Neonatal Intensive Care Unit to aid in the diagnosis and management of seizures. Neonatal seizures are particularly difficult to identify on the basis of clinical observation alone; diagnosis is greatly facilitated by CEEG monitoring. There is building evidence to suggest which neonates are at highest risk for seizures, and how CEEG can aid diagnosis. For the neurophysiologist, the unique features of neonatal seizures can distinguish them from nonictal patterns. These features include duration, location, morphology, and evolution. At the extreme, very frequent or prolonged neonatal seizures constitute status epilepticus. There is no consensus definition for neonatal status epilepticus, although the proposed criteria share some features. This article reviews available evidence to guide the application and interpretation of CEEG in the diagnosis of neonatal seizures and status epilepticus.

Monitoring neonatal seizures

Neonatal seizures are a neurological emergency and prompt treatment is required. Seizure burden in neonates can be very high, status epilepticus a frequent occurrence, and the majority of seizures do not have any clinical correlate. Detection of neonatal seizures is only possible with continuous electroencephalogram (EEG) monitoring. EEG interpretation requires special expertise that is not available in most neonatal intensive care units (NICUs). As a result, a simplified method of EEG recording incorporating an easy-to-interpret compressed trend of the EEG output (amplitude integrated EEG) from one of the EEG output from one or two channels has emerged as a popular way to monitor neurological function in the NICU. This is not without limitations; short duration and low amplitude seizures can be missed, artefacts are problematic and may mimic seizure-like activity and only a restricted area of the brain is monitored. Continuous multichannel EEG is the gold standard for detecting seizures and monitoring response to therapy but expert interpretation of the EEG output is generally not available. Some centres have set up remote access for neurophysiologists to the cot-side EEG, but reliable interpretation is wholly dependent on the 24 h availability of experts, an expensive solution. A more practical solution for the NICU without such expertise is an automated seizure detection system. This review outlines the current state of the art regarding cot-side monitoring of neonatal seizures in the NICU.

Robust neonatal EEG seizure detection through adaptive background modeling

Adaptive probabilistic modeling of the EEG background is proposed for seizure detection in neonates with hypoxic ischemic encephalopathy. The decision is made based on the temporal derivative of the seizure probability with respect to the adaptively modeled level of background activity. The robustness of the system to long duration "seizure-like" artifacts, in particular those due to respiration, is improved. The system was developed using statistical leave-one-patient-out performance assessment, on a large clinical dataset, comprising 38 patients of 1479 h total duration. The developed technique was then validated by a single test on a separate totally unseen randomized prospective dataset of 51 neonates totaling 2540 h of duration. By exploiting the proposed adaptation, the ROC area is increased from 93.4% to 96.1% (41% relative improvement). The number of false detections per hour is decreased from 0.42 to 0.24, while maintaining the correct detection of seizure burden at 70%. These results on the unseen data were predicted from the rigorous leave-one-patient-out validation and confirm the validity of our algorithm development process.

Neonatal seizures and status epilepticus

Neonatal seizures are common, often require EEG monitoring for diagnosis and management, may be associated with worse neurodevelopmental outcome, and can often be treated with existing anticonvulsants. A neonatal electrographic seizure is defined as a sudden, repetitive, evolving, and stereotyped event of abnormal electrographic pattern with amplitude of at least 2 μV and a minimum duration of 10 seconds. The diagnosis of neonatal seizures relies heavily on the neurophysiologist's interpretation of EEG. Consideration of specific criteria for the definition of a neonatal seizure, including seizure duration, location, morphology, evolution, semiology, and overall seizure burden, has utility for both the clinician and the researcher. The importance of EEG in the diagnosis and management of neonatal seizures, the electrographic characteristics of neonatal seizures, the impact of neonatal seizures on outcome, and tools to aid in the identification of neonatal seizures are reviewed.

Continuous electroencephalography monitoring in neonates

As more critically ill term and premature neonates are surviving their acute illness, their long-term neurodevelopmental morbidity is being recognized. Continuous monitoring of cerebral function, with electroencephalography or derived digital trends, can provide key information regarding seizures and background patterns, with direct treatment and prognostic implications. Conventional video-electroencephalography remains the gold standard for neonatal seizure diagnosis and quantification, but can be supplemented by digital trending modalities. Both conventional and amplitude-integrated electroencephalography can provide valuable data regarding the background trends. This review describes indications and methods for continuous electroencephalography monitoring in high-risk neonates.

Inclusion of temporal priors for automated neonatal EEG classification

The aim of this paper is to use recent advances in the clinical understanding of the temporal evolution of seizure burden in neonates with hypoxic ischemic encephalopathy to improve the performance of automated detection algorithms. Probabilistic weights are designed from temporal locations of neonatal seizure events relative to time of birth. These weights are obtained by fitting a skew-normal distribution to the temporal seizure density and introduced into the probabilistic framework of the previously developed neonatal seizure detector. The results are validated on the largest available clinical dataset, comprising 816.7 h. By exploiting these priors, the receiver operating characteristic area is increased by 23% (relative) reaching 96.74%. The number of false detections per hour is decreased from 0.45 to 0.25, while maintaining the correct detection of seizure burden at 70%.

Morphology-based automatic seizure detector for intracerebral EEG recordings

In this paper, a new seizure detection system aimed at assisting in a rapid review of prolonged intracerebral EEG recordings is described. It is based on quantifying the sharpness of the waveform, one of the most important electrographic EEG features utilized by experts for an accurate and reliable identification of a seizure. The waveform morphology is characterized by a measure of sharpness as defined by the slope of the half-waves. A train of abnormally sharp waves resulting from subsequent filtering are used to identify seizures. The method was optimized using 145 h of single-channel depth EEG from seven patients, and tested on another 158 h of single-channel depth EEG from another seven patients. Additionally, 725 h of depth EEG from 21 patients was utilized to assess the system performance in a multichannel configuration. Single-channel test data resulted in a sensitivity of 87% and a specificity of 71%. The multichannel test data reported a sensitivity of 81% and a specificity of 58.9%. The new system detected a wide range of seizure patterns that included rhythmic and nonrhythmic seizures of varying length, including those missed by the experts. We also compare the proposed system with a popular commercial system.

Model-based seizure detection for intracranial EEG recordings

This paper presents a novel model-based patient-specific method for automatic detection of seizures in the intracranial EEG recordings. The proposed method overcomes the complexities in the practical implementation of the patient-specific approach of seizure detection. The method builds a seizure model (set of basis functions) for a priori known seizure (the template seizure pattern), and uses the statistically optimal null filters as a building block for the detection of similar seizures. The process of modeling the template seizure is fully automatic. Overall, the detection method involves the segmentation of the template seizure pattern, rejection of the redundant and noisy segments, extraction of features from the segments to generate a set of models, selection of the best seizure model, and training of the classifier. The trained classifier is used to detect similar seizures in the remaining data. The resulting seizure detection method was evaluated on a total of 304 h of single-channel depth EEG recordings from 14 patients. The system performance is further compared to the Qu-Gotman patient-specific system using the same data. A significant improvement in the proposed system, in terms of specificity, is observed over the compared method.

Temporal evolution of seizure burden for automated neonatal EEG classification

The aim of this paper is to use recent advances in the clinical understanding of the temporal evolution of seizure burden in neonates with hypoxic ischemic encephalopathy to improve the performance of automated detection algorithms. Probabilistic weights are designed from temporal locations of neonatal seizure events relative to time of birth. These weights are obtained by fitting a skew-normal distribution to the temporal seizure density and introduced into the probabilistic framework of the previously developed neonatal seizure detector. The results are validated on the largest available clinical dataset, comprising 816.7 hours. By exploiting these priors, the ROC area is increased by 23% (relative) reaching 96.75%. The number of false detections per hour is decreased from 0.72 to 0.36, while maintaining the correct detection of seizure burden at 75%.

Instantaneous measure of EEG channel importance for improved patient-adaptive neonatal seizure detection

A measure of bipolar channel importance is proposed for EEG-based detection of neonatal seizures. The channel weights are computed based on the integrated synchrony of classifier probabilistic outputs for the channels which share a common electrode. These estimated time-varying weights are introduced within a Bayesian probabilistic framework to provide a channel specific and, thus, adaptive seizure classification scheme. Validation results on a clinical dataset of neonatal seizures confirm the utility of the proposed channel weighting for the two patient-independent seizure detectors recently developed by this research group: one based on support vector machines (SVMs) and the other on Gaussian mixture models (GMMs). By exploiting the channel weighting, the receiver operating characteristic (ROC) area can be significantly increased for the most difficult patients, with the average ROC area across 17 patients increased by 22% (relative) for the SVM and by 15% (relative) for the GMM-based detector, respectively. It is shown that the system developed here outperforms the recent published studies in this area.

Improving early seizure detection

Over the last decade, the search for a method able to reliably predict seizures hours in advance has been largely replaced by the more realistic goal of very early detection of seizure onset, which would allow therapeutic or warning devices to be triggered prior to the onset of disabling clinical symptoms. We explore in this article the steps along the pathway from data acquisition to closed-loop applications that can and should be considered to design the most efficient early seizure detection. Microelectrodes, high-frequency oscillations, high sampling rate, high-density arrays, and modern analysis techniques are all elements of the recording and detection process that in combination with modeling studies can provide new insights into the dynamics of seizure onsets. Each of these steps needs to be considered if detection devices that will favorably impact the quality of life of patients are to be implemented. This article is part of a Supplemental Special Issue entitled The Future of Automated Seizure Detection and Prediction.