A multistage knowledge-based system for EEG seizure detection in newborn infants

EEG signal description with spectral-envelope-based speech recognition features for detection of neonatal seizures

In this paper, features which are usually employed in automatic speech recognition (ASR) are used for the detection of seizures in newborn EEG. In particular, spectral envelope-based features, composed of spectral powers and their spectral derivatives are compared to the established feature set which has been previously developed for EEG analysis. The results indicate that the ASR features which model the spectral derivatives, either full-band or localized in frequency, yielded a performance improvement, in comparison to spectral-power-based features. Indeed it is shown here that they perform reasonably well in comparison with the conventional EEG feature set. The contribution of the ASR features was analyzed here using the support vector machines (SVM) recursive feature elimination technique. It is shown that the spectral derivative features consistently appear among the top-rank features. The study shows that the ASR features should be given a high priority when dealing with the description of the EEG signal.

Automatic detection of fast oscillations (40-200 Hz) in scalp EEG recordings

OBJECTIVE

We aim to automatically detect fast oscillations (40-200 Hz) related to epilepsy on scalp EEG recordings.

METHODS

The detector first finds localized increments of the signal power in narrow frequency bands. A simple classification based on two features, a narrowband to wideband signal amplitude ratio and an absolute narrowband signal amplitude, then allows for an important reduction in the number of false positives.

RESULTS

When compared to an expert, the performance in 15 focal epilepsy patients resulted in 3.6 false positives per minute at 95% sensitivity, with at least 40% of the detected events being true positives. In most of the patients the channels showing the highest number of events according to the expert and the automatic detector were the same.

CONCLUSIONS

A high sensitivity is achieved with the proposed automatic detector, but results should be reviewed by an expert to remove false positives.

SIGNIFICANCE

The time required to mark fast oscillations on scalp EEG recordings is drastically reduced with the use of the proposed detector. Thus, the automatic detector is a useful tool in studies aiming to create a better understanding of the fast oscillations visible on the scalp.

Speech recognition features for EEG signal description in detection of neonatal seizures

In this work, features which are usually employed in automatic speech recognition (ASR) are used for the detection of neonatal seizures in newborn EEG. Three conventional ASR feature sets are compared to the feature set which has been previously developed for this task. The results indicate that the thoroughly-studied spectral envelope based ASR features perform reasonably well on their own. Additionally, the SVM Recursive Feature Elimination routine is applied to all extracted features pooled together. It is shown that ASR features consistently appear among the top-rank features.

Performance assessment for EEG-based neonatal seizure detectors

Objective

This study discusses an appropriate framework to measure system performance for the task of neonatal seizure detection using EEG. The framework is used to present an extended overview of a multi-channel patient-independent neonatal seizure detection system based on the Support Vector Machine (SVM) classifier.

Methods

The appropriate framework for performance assessment of neonatal seizure detectors is discussed in terms of metrics, experimental setups, and testing protocols. The neonatal seizure detection system is evaluated in this framework. Several epoch-based and event-based metrics are calculated and curves of performance are reported. A new metric to measure the average duration of a false detection is proposed to accompany the event-based metrics. A machine learning algorithm (SVM) is used as a classifier to discriminate between seizure and non-seizure EEG epochs. Two post-processing steps proposed to increase temporal precision and robustness of the system are investigated and their influence on various metrics is shown. The resulting system is validated on a large clinical dataset of 267 h.

Results

In this paper, it is shown how a complete set of metrics and a specific testing protocol are necessary to extensively describe neonatal seizure detection systems, objectively assess their performance and enable comparison with existing alternatives. The developed system currently represents the best published performance to date with an ROC area of 96.3%. The sensitivity and specificity were ∼90% at the equal error rate point. The system was able to achieve an average good detection rate of ∼89% at a cost of 1 false detection per hour with an average false detection duration of 2.7 min.

Conclusions

It is shown that to accurately assess the performance of EEG-based neonatal seizure detectors and to facilitate comparison with existing alternatives, several metrics should be reported and a specific testing protocol should be followed. It is also shown that reporting only event-based metrics can be misleading as they do not always reflect the true performance of the system.

Significance

This is the first study to present a thorough method for performance assessment of EEG-based seizure detection systems. The evaluated SVM-based seizure detection system can greatly assist clinical staff, in a neonatal intensive care unit, to interpret the EEG.

EEG-based neonatal seizure detection with Support Vector Machines

OBJECTIVE

The study presents a multi-channel patient-independent neonatal seizure detection system based on the Support Vector Machine (SVM) classifier.

METHODS

A machine learning algorithm (SVM) is used as a classifier to discriminate between seizure and non-seizure EEG epochs. Two post-processing steps are proposed to increase both the temporal precision and the robustness of the system. The resulting system is validated on a large clinical dataset of 267 h of EEG data from 17 full-term newborns with seizures.

RESULTS

The performance of the system using event-based metrics is reported. The system showed the best up-to-date performance of a neonatal seizure detection system. The system was able to achieve an average good detection rate of ~89% with one false seizure detection per hour, ~96% with two false detections per hour, or ~100% with four false detections per hour. An analysis of errors revealed sources of misclassification in terms of both missed seizures and false detections.

CONCLUSIONS

The results obtained with the proposed SVM-based seizure detection system allow for its practical application in neonatal intensive care units.

SIGNIFICANCE

The proposed SVM-based seizure detection system can greatly assist clinical staff, in a neonatal intensive care unit, to interpret the EEG. The system allows control of the final decision by choosing different confidence levels which makes it flexible for clinical needs. The obtained results may provide a reference for future seizure detection systems.

An international survey of EEG use in the neonatal intensive care unit

OBJECTIVE

To examine the extent of EEG monitoring in neonatal intensive care units (NICUs), and to survey the level of experience and training of those using it.

STUDY DESIGN

A web-based survey, the link to which was circulated via e-mail, personal contact, specialist societies and professional groups. Survey data were exported to SPSS for analysis.

RESULTS

In total 210 surveys were analysed; 124 from Europe, 54 from the US. Ninety percent of respondents had access to either EEG or aEEG monitoring; 51% had both. EEG was mainly interpreted by neurophysiologists (72%) whereas aEEG was usually interpreted by neonatologists (80%). Only 9% of respondents reported that they felt 'very confident' in their ability to interpret aEEG/EEG with 31% reporting that they were 'not confident'. Half had received no formal training in EEG.

CONCLUSION

Both aEEG and conventional EEG were used extensively in the NICUs surveyed for this study. Most of the survey respondents were not confident in their ability to interpret EEGs despite the fact that they used monitoring routinely. There is an urgent need for a structured and appropriately targeted training programme in EEG methodologies and EEG interpretation for neonatal intensive care unit staff.

Gaussian mixture models for classification of neonatal seizures using EEG

A real-time neonatal seizure detection system is proposed based on a Gaussian mixture model classifier. The system includes feature transformation techniques and classifier output postprocessing. The detector was evaluated on a database of 20 patients with 330 h of recordings. A detailed analysis of the choice of parameters for the detector is provided. A mean good detection rate of 79% was obtained with only 0.5 false detections per hour. A thorough review of all misclassified events was performed, from which a number of patterns causing false detections were identified.

High-yield decomposition of surface EMG signals

OBJECTIVE

Automatic decomposition of surface electromyographic (sEMG) signals into their constituent motor unit action potential trains (MUAPTs).

METHODS

A small five-pin sensor provides four channels of sEMG signals that are in turn processed by an enhanced artificial intelligence algorithm evolved from a previous proof-of-principle. We tested the technology on sEMG signals from five muscles contracting isometrically at force levels ranging up to 100% of their maximal level, including those that were covered with more than 1.5cm of adipose tissue. Decomposition accuracy was measured by a new method wherein a signal is first decomposed and then reconstructed and the accuracy is measured by comparison. Results were confirmed by the more established two-source method.

RESULTS

The number of MUAPTs decomposed varied among muscles and force levels and mostly ranged from 20 to 30, and occasionally up to 40. The accuracy of all the firings of the MUAPTs was on average 92.5%, at times reaching 97%.

CONCLUSIONS

Reported technology can reliably perform high-yield decomposition of sEMG signals for isometric contractions up to maximal force levels.

SIGNIFICANCE

The small sensor size and the high yield and accuracy of the decomposition should render this technology useful for motor control studies and clinical investigations.

Seizure detection in intracranial EEG using a fuzzy inference system

In this paper, we present a fuzzy rule-based system for the automatic detection of seizures in the intracranial EEG (IEEG) recordings. A total of 302.7 hours of the IEEG with 78 seizures, recorded from 21 patients aged between 10 and 47 years were used for the evaluation of the system. After preprocessing, temporal, spectral, and complexity features were extracted from the segmented IEEGs. The results were thresholded using the statistics of a reference window and integrated spatio-temporally using a fuzzy rule-based decision making system. The system yielded a sensitivity of 98.7%, a false detection rate of 0.27/h, and an average detection latency of 11 s. The results from the automatic system correlate well with the visual analysis of the seizures by the expert. This system may serve as a good seizure detection tool for monitoring long-term IEEG with relatively high sensitivity and low false detection rate.

An SVM-based system and its performance for detection of seizures in neonates

This work presents a multi-channel patient-independent neonatal seizure detection system based on the SVM classifier. Several post-processing steps are proposed to increase temporal precision and robustness of the system and their influence on performance is shown. The SVM-based system is evaluated on a large clinical dataset using several epoch-based and event based metrics and curves of performance are reported. Additionally, a new metric to measure the average duration of a false detection is proposed to accompany the event-based metrics.

Amplitude-integrated electroencephalography in neonates

Conventional electroencephalography (EEG) has been used for decades in the neonatal intensive care unit for formulating neurologic prognoses, demonstrating brain functional state and degree of maturation, revealing cerebral lesions, and identifying the presence and number of electrographic seizures. However, both the immediate availability of conventional EEG and the expertise with which it is interpreted are variable. Amplitude-integrated EEG provides simplified monitoring of cerebral function, and is rapidly gaining popularity among neonatologists, with growing use in bedside decision making and inclusion criteria for randomized clinical studies. Nonetheless, child neurologists and neurophysiologists remain cautious about relying solely on this tool and prefer interpreting conventional EEG. The present review examines the technical aspects of generating, recording, and interpreting amplitude-integrated EEG and contrasts this approach with conventional EEG. Finally, several proposed amplitude-integrated EEG classification schemes are reviewed. A clear understanding of this emerging technology of measuring brain health in the premature or sick neonate is critical in modern care of the newborn infant.

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

This paper describes the design and test results of a three-stage automated system for neonatal EEG seizure detection. Stage I of the system is the initial detection stage and identifies overlapping 5-second segments of suspected seizure activity in each EEG channel. In stage II, the detected segments from stage I are spatiotemporally clustered to produce multichannel candidate seizures. In stage III, the candidate seizures are processed further using measures of quality and context-based rules to eliminate false candidates. False candidates because of artifacts and commonly occurring EEG background patterns such as bifrontal delta activity are also rejected. Seizures at least 10 seconds in duration are considered for reporting results. The testing data consisted of recordings of 28 seizure subjects (34 hours of data) and 48 nonseizure subjects (87 hours of data) obtained in the neonatal intensive care unit. The data were not edited to remove artifacts and were identical in every way to data normally processed visually. The system was able to detect seizures of widely varying morphology with an average detection sensitivity of almost 80% and a subject sensitivity of 96%, in comparison with a team of clinical neurophysiologists who had scored the same recordings. The average false detection rate obtained in nonseizure subjects was 0.74 per hour.

A fuzzy rule-based system for epileptic seizure detection in intracranial EEG

OBJECTIVE

We present a method for automatic detection of seizures in intracranial EEG recordings from patients suffering from medically intractable focal epilepsy.

METHODS

We designed a fuzzy rule-based seizure detection system based on knowledge obtained from experts' reasoning. Temporal, spectral, and complexity features were extracted from IEEG segments, and spatio-temporally integrated using the fuzzy rule-based system for seizure detection. A total of 302.7h of intracranial EEG recordings from 21 patients having 78 seizures was used for evaluation of the system.

RESULTS

The system yielded a sensitivity of 98.7%, a false detection rate of 0.27/h, and an average detection latency of 11s. There was only one missed seizure. Most of false detections were caused by high-amplitude rhythmic activities. The results from the system correlate well with those from expert visual analysis.

CONCLUSION

The fuzzy rule-based seizure detection system enabled us to deal with imprecise boundaries between interictal and ictal IEEG patterns.

SIGNIFICANCE

This system may serve as a good seizure detection tool with high sensitivity and low false detection rate for monitoring long-term IEEG.

A semi-automated method for epileptiform transient detection in the EEG of the fetal sheep using time-frequency analysis

Perinatal hypoxia remains a significant cause of brain damage. Currently there are no biomarkers to detect the at risk brain. Recent research, however, suggests that the appearance of epileptiform transients in the first 6-8 hours after hypoxia (the latent phase of injury) are predictive of neural outcome. To quantify this further a key need is to automate EEG signal analysis to aid clinical staff with the vast amounts of complex data to review. In this study, we present a semi-automated method for spike detection in the fetal sheep EEG. The method utilizes the short time Fourier transform and peak separation to extract spikes. The performance of the method was found to be high in sensitivity and selectivity over 3 distinct time points.

Identifying airway obstructions using photoplethysmography (PPG)

OBJECTIVE

Central and obstructive apneas are sources of morbidity and mortality associated with primary patient conditions as well as secondary to medical care such as sedation/analgesia in post-operative patients. This research investigates the predictive value of the respirophasic variation in the noninvasive photoplethysmography (PPG) waveform signal in detecting airway obstruction.

METHODS

PPG data from 20 consenting healthy adults (12 male, 8 female) undergoing anesthesia were collected directly after surgery and before transfer to the Post Anesthesia Care Unit (PACU). Features of the PPG waveform were calculated and used in a neural network to classify normal and obstructive events.

RESULTS

During the postoperative period studied, the neural network classifier yielded an average (+/-standard deviation) 75.4 (+/-3.7)% sensitivity, 91.6 (+/-2.3)% specificity, 84.7 (+/-3.5)% positive predictive value, 85.9 (+/-1.8)% negative predictive value, and an overall accuracy of 85.4 (+/-2.0)%.

CONCLUSIONS

The accuracy of this method shows promise for use in real-time monitoring situations.

Seizure detection in neonates: Improved classification through supervised adaptation

The goal of neonatal seizure detection is the development of a patient independent system to alert staff in the neonatal intensive care unit of ongoing seizures. This study demonstrates the potential in adapting a patient independent classifier using patient specific data. Supervised adaptation is investigated using the basic gradient descent algorithm and least mean squares procedures. An increase in mean ROC area of 3% is obtained for the best performing learning algorithm, yielding an increase in mean accuracy of 7.7% compared to the patient independent algorithm.

Automated single channel seizure detection in the neonate

Neonatal seizures are the most common neurological emergency in the neonatal period and are associated with poor long-term outcome. EEG is considered the gold standard for identification of all neonatal seizures, reducing the number of EEG electrodes required would reduce patient handling and allow faster acquisition of data. A method for automated neonatal seizure detection based on two carefully chosen cerebral scalp electrodes but trained using multi-channel EEG is presented. The algorithm was developed and tested using a multi-channel EEG dataset containing 411 seizures from 251.9 hours of EEG recorded from 17 full-term neonates. Automated seizure detection using a variety of bipolar channel derivations was investigated. Channel C3-C4 yielded correct detection of 90.77% of seizures with a false detection rate of 9.43%. This compares favourably with a multi-channel seizure detection method which detected 81.03% of seizures with a false detection rate of 3.82%.