Seizure Prediction Algorithm Research Articles

A signal regularity-based automated seizure prediction algorithm using long-term scalp EEG recordings

The purpose of this study was to evaluate a signal regularity-based automated seizure prediction algorithm for scalp EEG. Signal regularity was quantified using the Pattern Match Regularity Statistic (PMRS), a statistical measure. The primary feature of the prediction algorithm is the degree of convergence in PMRS (“PMRS entrainment”) among the electrode groups determined in the algorithm training process. The hypothesis is that the PMRS entrainment increases during the transition between interictal and ictal states, and therefore may serve as an indicator for prediction of an impending seizure.

Seizure prediction with spectral power of EEG using cost-sensitive support vector machines

We propose a patient-specific algorithm for seizure prediction using multiple features of spectral power from electroencephalogram (EEG) and support vector machine (SVM) classification. The proposed patient-specific algorithm consists of preprocessing, feature extraction, SVM classification, and postprocessing. Preprocessing removes artifacts of intracranial EEG recordings and they are further preprocessed in bipolar and/or time-differential methods. Features of spectral power of raw, or bipolar and/or time-differential intracranial EEG (iEEG) recordings in nine bands are extracted from a sliding 20-s-long and half-overlapped window. Nine bands are selected based on standard EEG frequency bands, but the wide gamma bands are split into four. Cost-sensitive SVMs are used for classification of preictal and interictal samples, and double cross-validation is used to achieve in-sample optimization and out-of-sample testing. We postprocess SVM classification outputs using the Kalman Filter and it removes sporadic and isolated false alarms. The algorithm has been tested on iEEG of 18 patients of 20 available in the Freiburg EEG database who had three or more seizure events. To investigate the discriminability of the features between preictal and interictal, we use the Kernel Fisher Discriminant analysis. The proposed patient-specific algorithm for seizure prediction has achieved high sensitivity of 97.5% with total 80 seizure events and a low false alarm rate of 0.27 per hour and total false prediction times of 13.0% over a total of 433.2 interictal hours by bipolar preprocessing (92.5% sensitivity, a false positive rate of 0.20 per hour, and false prediction times of 9.5% by time-differential preprocessing). This high prediction rate demonstrates that seizures can be predicted by the patient-specific approach using linear features of spectral power and nonlinear classifiers. Bipolar and/or time-differential preprocessing significantly improves sensitivity and specificity. Spectral powers in high gamma bands are the most discriminating features between preictal and interictal. High sensitivity and specificity are achieved by nonlinear classification of linear features of spectral power. Power changes in certain frequency bands already demonstrated their possibilities for seizure prediction indicators, but we have demonstrated that combining those spectral power features and classifying them in a multivariate approach led to much higher prediction rates. Employing only linear features is advantageous, especially when it comes to an implantable device, because they can be computed rapidly with low power consumption.

EFFECT OF STIMULUS PARAMETERS IN THE TREATMENT OF SEIZURES BY ELECTRICAL STIMULATION IN THE KAINATE ANIMAL MODEL

Preliminary results from animal and clinical studies demonstrate that electrical stimulation of brain structures can reduce seizure frequency in patients with refractory epilepsy. Since most researchers derive stimulation parameters by trial and error, it is unclear what stimulation frequency, amplitude and duration constitutes a set of optimal stimulation parameters for aborting seizure activity in a given patient. In this investigation, we begin to quantify the independent effects of stimulation parameters on electrographic seizures, such that they could be used to develop an efficient closed-loop prosthesis that intervenes before the clinical onset of a seizure and seizure generalization. Biphasic stimulation is manually delivered to the hippocampus in response to a visually detected electrographic seizure. Such focal, responsive stimulation allows for anti-seizure treatment delivery with improved temporal and spatial specificity over conventional open-loop stimulation paradigms, with the possibility of avoiding tissue damage stemming from excessive exposure to electrical stimulation. We retrospectively examine the effects of stimulation frequency (low, medium and high), pulse-width (low and high) and amplitude (low and high) in seizures recorded from 23 kainic acid treated rats. We also consider the effects of total charge delivered and the rate of charge delivery, and identify stimulation parameter sets that induce after-discharges or more seizures. Among the stimulation parameters evaluated, we note 2 major findings. First, stimulation frequency is a key parameter for inhibiting seizure activity; the anti-seizure effect cannot be attributed to only the charge delivered per phase. Second, an after-discharge curve shows that as the frequency and pulse-width of stimulation increases, smaller pulse amplitudes are capable of eliciting an after-discharge. It is expected that stimulation parameter optimization will lead to devices with enhanced treatment efficacies and reduced side-effect profiles, especially when used in conjunction with seizure prediction or detection algorithms in a closed-loop control application.

A brief survey of computational models of normal and epileptic EEG signals: A guideline to model-based seizure prediction

In recent decades, seizure prediction has caused a lot of research in both signal processing and the neuroscience field. The researches have tried to enhance the conventional seizure prediction algorithms such that the rate of the false alarms be appropriately small, so that seizures can be predicted according to clinical standards. To date, none of the proposed algorithms have been sufficiently adequate. In this article we show that in considering the mechanism of the generation of seizures, the prediction results may be improved. For this purpose, an algorithm based on the identification of the parameters of a physiological model of seizures is introduced. Some models of electroencephalographic (EEG) signals that can also be potentially considered as models of seizure and some developed seizure models are reviewed. As an example the model of depth-EEG signals, proposed by Wendling, is studied and is shown to be a suitable model.

EPILEPSIAE – A European epilepsy database

With a worldwide prevalence of about 1%, epilepsy is one of the most common serious brain diseases with profound physical, psychological and, social consequences. Characteristic symptoms are seizures caused by abnormally synchronized neuronal activity that can lead to temporary impairments of motor functions, perception, speech, memory or, consciousness.The possibility to predict the occurrence of epileptic seizures by monitoring the electroencephalographic activity (EEG) is considered one of the most promising options to establish new therapeutic strategies for the considerable fraction of patients with currently insufficiently controlled seizures.Here, a database is presented which is part of an EU-funded project “EPILEPSIAE” aiming at the development of seizure prediction algorithms which can monitor the EEG for seizure precursors. High-quality, long-term continuous EEG data, enriched with clinical metadata, which so far have not been available, are managed in this database as a joint effort of epilepsy centers in Portugal (Coimbra), France (Paris) and Germany (Freiburg).The architecture and the underlying schema are here reported for this database. It was designed for an efficient organization, access and search of the data of 300 epilepsy patients, including high quality long-term EEG recordings, obtained with scalp and intracranial electrodes, as well as derived features and supplementary clinical and imaging data. The organization of this European database will allow for accessibility by a wide spectrum of research groups and may serve as a model for similar databases planned for the future.

Out of (Dis)order? the Dynamics of Seizure Initiation

Network Dynamics During Development of Pharmacologically Induced Epileptic Seizures in Rats in Vivo. Cymerblit-Sabba A, Schiller Y. J Neurosci 2010;30(5):1619–1630. In epilepsy, the cortical network fluctuates between the asymptomatic interictal state and the symptomatic ictal state of seizures. Despite their importance, the network dynamics responsible for the transition between the interictal and ictal states are largely unknown. Here we used multielectrode single-unit recordings from the hippocampus to investigate the network dynamics during the development of seizures evoked by various chemoconvulsants in vivo. In these experiments, we detected a typical network dynamics signature that preceded seizure initiation. The preictal state preceding pilocarpine-, kainate-, and picrotoxin-induced seizures was characterized by biphasic network dynamics composed of an early desynchronization phase in which the tendency of neurons to fire correlated action potentials decreased, followed by a late resynchronization phase in which the activity and synchronization of the network gradually increased. This biphasic network dynamics preceded the initiation both of the initial seizure and of recurrent spontaneous seizures that followed. During seizures, firing of individual neurons and interneuronal synchronization further increased. These findings advance our understanding of the network dynamics leading to seizure initiation and may in future help in the development of novel seizure prediction algorithms.

Seizure Prediction With Spectral Power of EEG Using Cost-Sensitive Support Vector Machines

A patient-specific seizure prediction algorithm is proposed using a classifier to differentiate pre-ictal from inter-ictal EEG signals. The spectral power of EEG processed in four different fashions is used as features: raw, time-differential, space-differential, and time/space-differential EEG. The features are classified using cost-sensitive support vector machines by the double cross-validation methodology. The proposed algorithm has been applied to EEG recordings of 18 patients in the Freiburg EEG database, totaling 80 seizures and 437 h long inter-ictal recordings. Classification with the feature obtained from time/space-differential ECoG demonstrates the performance of 86.25% sensitivity and 0.1281 false positives per hour in out-of-sample testing.

Network Dynamics during Development of Pharmacologically Induced Epileptic Seizures in RatsIn Vivo

In epilepsy, the cortical network fluctuates between the asymptomatic interictal state and the symptomatic ictal state of seizures. Despite their importance, the network dynamics responsible for the transition between the interictal and ictal states are largely unknown. Here we used multielectrode single-unit recordings from the hippocampus to investigate the network dynamics during the development of seizures evoked by various chemoconvulsants in vivo. In these experiments, we detected a typical network dynamics signature that preceded seizure initiation. The preictal state preceding pilocarpine-, kainate-, and picrotoxin-induced seizures was characterized by biphasic network dynamics composed of an early desynchronization phase in which the tendency of neurons to fire correlated action potentials decreased, followed by a late resynchronization phase in which the activity and synchronization of the network gradually increased. This biphasic network dynamics preceded the initiation both of the initial seizure and of recurrent spontaneous seizures that followed. During seizures, firing of individual neurons and interneuronal synchronization further increased. These findings advance our understanding of the network dynamics leading to seizure initiation and may in future help in the development of novel seizure prediction algorithms.

A common strategy and database to compare the performance of seizure prediction algorithms

A reliable algorithm for the timely prediction of epileptic seizures would be a milestone in epilepsy research. Prediction performances have so far been determined using retrospective data assessment, leaving open the question as to whether they prove statistically significant and clinically useful under prospective conditions. To this aim, a Seizure Prediction Competition has been set up. Here, the background and the details of this competition are described.

Comments on: Optimization and data mining in biomedicine

In this review paper, Pardalos et al. discuss selected applications of operation research and system analysis in biomedicine. Some of the results presented in this paper, such as consistent biclustering and seizure prediction algorithms, are developed by Pardalos and his research group. As for the material included in each of the sections, I believe that it is an excellent primer for young researchers who seek for introductory literature on applying optimization to solving problems in biomedicine. The part on radiation therapy introduces the background on the application and the underlying optimization techniques with references to the medical instrumentation. Then models and solution approaches are presented and discussed. The theoretical results discussed by Busygin et al. for consistent biclustering are of particular importance since, in general, there are not many papers discussing the theoretical foundations of biclustering. The third application presents an alternative and innovative use of operations research in the electrode selection problem of EEG for improving the epileptic seizure prediction algorithm based on nonlinear dynamic time series analysis. This application shows how a formulation that is typically applied to problems of industrial interest can be modified and applied in a totally new framework. The inclusion of EEG source localization problem in this survey is also a nice addition to the paper since the algorithms developed in this field are mainly based on signal processing approaches. An operations research point of view may give some more insight and new ideas for this inverse problem. This paper is a good reference paper for operational researchers and mathematicians and I believe it will be well received by the readers of TOP.

Seizure prediction: Any better than chance?

To test whether epileptic seizure prediction algorithms have true predictive power, their performance must be compared with the one expected under well-defined null hypotheses. For this purpose, analytical performance estimates and seizure predictor surrogates were introduced. We here extend the Monte Carlo framework of seizure predictor surrogates by introducing alarm times surrogates. We construct artificial seizure time sequences and artificial seizure predictors to be consistent or inconsistent with various null hypotheses to determine the frequency of null hypothesis rejections obtained from analytical performance estimates and alarm times surrogates under controlled conditions. Compared to analytical performance estimates, alarm times surrogates are more flexible with regard to the testable null hypotheses. Both approaches have similar, high statistical power to indicate true predictive power. For Poisson predictors that fulfill the null hypothesis of analytical performance estimates, the frequency of false positive null hypothesis rejections can exceed the significance level for long mean inter-alarm intervals, revealing an intrinsic bias of these analytical estimates. Alarm times surrogates offer important advantages over analytical performance estimates. The key question in the field of seizure prediction is whether seizures can in principle be predicted or whether algorithms which have been presumed to perform better than chance actually are unable to predict seizures and simply have not yet been tested against the appropriate null hypotheses. Alarm times surrogates can help to answer this question.

The statistics of a practical seizure warning system

Statistical methods for evaluating seizure prediction algorithms are controversial and a primary barrier to realizing clinical applications. Experts agree that these algorithms must, at a minimum, perform better than chance, but the proper method for comparing to chance is in debate. We derive a statistical framework for this comparison, the expected performance of a chance predictor according to a predefined scoring rule, which is in turn used as the control in a hypothesis test. We verify the expected performance of chance prediction using Monte Carlo simulations that generate random, simulated seizure warnings of variable duration. We propose a new test metric, the difference between algorithm and chance sensitivities given a constraint on proportion of time spent in warning, and use a simple spectral power-based measure to demonstrate the utility of the metric in four patients undergoing intracranial EEG monitoring during evaluation for epilepsy surgery. The methods are broadly applicable to other scoring rules. We present them as an advance in the statistical evaluation of a practical seizure advisory system.

Seizure Prediction

There is mounting evidence that seizures are preceded by characteristic changes in the EEG that are detectable minutes before seizure onset. Using novel signal analysis techniques, researchers are beginning to characterize the transition from the interictal to the ictal state in quantitative terms. This research has led to the development of automated seizure prediction algorithms. Active debate persists regarding the interpretation of research results, methods of signal analysis, as well as experimental and statistical methods for testing seizure prediction algorithms. Developments in this field have led to new theories on the mechanism of seizure development and resolution. The ability to predict seizures could lead the way to novel diagnostic and therapeutic methods for the treatment of patients with epilepsy.

Do False Predictions of Seizures Depend on the State of Vigilance? A Report from Two Seizure‐Prediction Methods and Proposed Remedies

Available seizure-prediction algorithms are accompanied by high numbers of false predictions to achieve high sensitivity. Little is known about the extent to which changes in EEG dynamics contribute to false predictions. This study addresses potential causes and the circadian distribution of false predictions as well as their relation to the sleep-wake cycle. In 21 patients, each with 24 h of interictal invasive EEG recordings, two methods, the dynamic similarity index and the mean phase coherence, were assessed with respect to time points of false predictions. Visual inspection of the invasive EEG data and additional scalp electroencephalogram data was performed at times of false predictions to identify possible correlates of changes in the EEG dynamics. A dependency of false predictions on the time of day is shown. Renormalized to the duration of the period patients are asleep and awake, 86% of all false predictions occurred during sleep for the dynamic similarity index and 68% for the mean phase coherence, respectively. Combining two reference intervals, one during sleep and one in an awake state, the dynamic similarity index increases its performance by reducing the number of false predictions by almost 50% without major changes in sensitivity. No obvious dependence of false predictions was noted on visible epileptic activity, such as spikes, sharp waves, or subclinical ictal patterns. Changes in the EEG dynamics related to the sleep-wake cycle contribute to limits of specificity of both seizure-prediction methods investigated. This may provide a clue for improving prediction methods in general. The combination of reference states yields promising results and may offer opportunities to increase further the performance of prediction methods.

An epileptic seizure prediction algorithm based on second-order complexity measure

The quality of life of many epilepsy patients may be improved significantly if the occurrence of epileptic seizures can be successfully forecasted and clinical intervention, such as electrical stimulation or drug delivery, can then be used to suppress their emergence, or warn the patient of the forthcoming events. In this paper, a prediction algorithm based on the second-order complexity measure was proposed to predict the impending seizures. Through the analysis of long-term intracranial EEG recordings from two frontal lobe epilepsy patients, the results indicated that the sensitivity of prediction was 77.8% (14/18) and 66.7% (4/6) and the number of false warnings was 3 and 2 for the two patients, respectively. Because only the information of past seizures was utilized to predict the current seizure and the computation load was low, the prediction algorithm could possibly be applied to clinical practice.

Long-term prospective on-line real-time seizure prediction

Objective Epilepsy, one of the most common neurological disorders, constitutes a unique opportunity to study the dynamics of spatiotemporal state transitions in real, complex, nonlinear dynamical systems. In this study, we evaluate the performance of a prospective on-line real-time seizure prediction algorithm in two patients from a common database. Methods We previously demonstrated that measures of chaos and angular frequency, estimated from electroencephalographic (EEG) signals recorded at critical sites in the cerebral cortex, progressively converge (i.e. become dynamically entrained) as the epileptic brain transits from the asymptomatic interictal state to the ictal state (seizure) (Iasemidis et al., 2001, 2002a, 2003a). This observation suggested the possibility of developing algorithms to predict seizures well ahead of their occurrences. One of the central points in those investigations was the application of optimization theory, specifically quadratic zero-one programming, for the selection of the critical cortical sites. This current study combines that observation with a dynamical entrainment detection method to prospectively predict epileptic seizures. The algorithm was tested in two patients with long-term (107.54 h) and multi-seizure EEG data B and C (Lehnertz and Litt, 2004). Results Analysis from the 2 test patients resulted in the prediction of up to 91.3% of the impending 23 seizures, about 89±15 min prior to seizure onset, with an average false warning rate of one every 8.27 h and an allowable prediction horizon of 3 h. Conclusions The algorithm provides warning of impending seizures prospectively and in real time, that is, it constitutes an on-line and real-time seizure prediction scheme. Significance These results suggest that the proposed seizure prediction algorithm could be used in novel diagnostic and therapeutic applications in epileptic patients.

Clinical neurophysiology of epilepsy

Electroencephalography (EEG) continues to be the most important diagnostic tool in the management of patients with epilepsy. In particular, the high specificity of interictal epileptiform discharges makes scalp EEG a valuable tool in the evaluation of patients with a history of seizures or seizure-like episodes. Advances in technology, most notably the development of digital video-EEG, have significantly expanded the utility of EEG. In addition to the routine EEG, long-term monitoring studies including video-EEG, ambulatory EEG, and continuous EEG monitoring play important roles in various aspects of the diagnosis and treatment of epilepsy. Recent developments in computerized seizure detection and prediction algorithms, particularly those utilizing intracranial EEG electrodes, hold promise for future development of novel treatment strategies.

Prediction of PTZ-induced seizures using wavelet-based residual entropy of cortical and subcortical field potentials.

Our proposed algorithm for seizure prediction is based on the principle that seizure build-up is always preceded by constantly changing bursting levels. We use a novel measure of residual subband wavelet entropy (RSWE) to directly estimate the entropy of bursts, which is otherwise obscured by the ongoing background activity. Our results are obtained using a slow infusion anesthetized pentylenetetrazol (PTZ) rat model in which we record field potentials (FPs) from frontal cortex and two thalamic areas (anterior and posterior nuclei). In each frequency band, except for the theta-delta frequency bands, we observed a significant build-up of RSWE from the preictal period to the first ictal event (p < or = 0.05) in cortex. Significant differences were observed between cortical and thalamic RSWE (p < or = 0.05) subsequent to seizure development. A key observation is the twofold increase in mean cortical RSWE from the preictal to interictal period. Exploiting this increase, we develop a slope change detector to discern early acceleration of entropy and predict the approaching seizure. We use multiple observations through sequential detection of slope changes to enhance the sensitivity of our prediction. Using the proposed method applied to a cohort of four rats subjected to PTZ infusion, we were able to predict the first seizure episode 28 min prior to its occurrence.

Prediction of Human Epileptic Seizures based on Optimization and Phase Changes of Brain Electrical Activity

The phenomenon of epilepsy, one of the most common neurological disorders, constitutes a unique opportunity to study the dynamics of spatiotemporal state transitions in real, complex, nonlinear dynamical systems. We previously demonstrated that measures of chaos and angular frequency obtained from electroencephalographic (EEG) signals generated by critical sites in the cerebral cortex converge progressively (dynamical entrainment) from the asymptomatic interictal state to the ictal state (seizure) [L.D. Iasemidis, P. Pardalos, J.C. Sackellares and D.-S. Shiau (2001). Quadratic binary programming and dynamical system approach to determine the predictability of epileptic seizures. J. Combinatorial Optimization, 5, 9–26; L.D. Iasemidis, D.-S. Shiau, P.M. Pardalos and J.C. Sackellares (2002). Phase entrainment and predictability of epileptic seizures. In: P.M. Pardalos and J. Principe (Eds.), Biocomputing, pp. 59–84. Kluwer Academic Publishers]. This observation suggests the possibility of developing algorithms to predict seizures. One of the central points of those investigations was the application of optimization theory, specifically quadratic zero-one programming, for the selection of the cortical sites that exhibit preictal dynamical entrainment. In this study we present results from the application of this methodology to the prediction of epileptic seizures. Analysis of continuous, long-term (18–140 h), multielectrode EEG recordings from 5 patients resulted in the prediction of 88% of the impending 50 seizures, on average about 83 min prior to seizure onset, with an average false warning rate of one every 5.26 h. These results suggest that this seizure prediction algorithm performs well enough to be used in diagnostic and therapeutic applications in epileptic patients. Similar algorithms may be useful for certain spatiotemporal state transitions in other physical and biological systems.