Comparison of predictability of epileptic seizures by a linear and a nonlinear method

Advanced framework for epilepsy detection through image-based EEG signal analysis

Background

Recurrent and unpredictable seizures characterize epilepsy, a neurological disorder affecting millions worldwide. Epilepsy diagnosis is crucial for timely treatment and better outcomes. Electroencephalography (EEG) time-series data analysis is essential for epilepsy diagnosis and surveillance. Complex signal processing methods used in traditional EEG analysis are computationally demanding and difficult to generalize across patients. Researchers are using machine learning to improve epilepsy detection, particularly visual feature extraction from EEG time-series data.

Objective

This study examines the application of a Gramian Angular Summation Field (GASF) approach for the analysis of EEG signals. Additionally, it explores the utilization of image features, specifically the Scale-Invariant Feature Transform (SIFT) and Oriented FAST and Rotated BRIEF (ORB) techniques, for the purpose of epilepsy detection in EEG data.

Methods

The proposed methodology encompasses the transformation of EEG signals into images based on GASF, followed by the extraction of features utilizing SIFT and ORB techniques, and ultimately, the selection of relevant features. A state-of-the-art machine learning classifier is employed to classify GASF images into two categories: normal EEG patterns and focal EEG patterns. Bern-Barcelona EEG recordings were used to test the proposed method.

Results

This method classifies EEG signals with 96% accuracy using SIFT features and 94% using ORB features. The Random Forest (RF) classifier surpasses state-of-the-art approaches in precision, recall, F1-score, specificity, and Area Under Curve (AUC). The Receiver Operating Characteristic (ROC) curve shows that Random Forest outperforms Support Vector Machine (SVM) and k-Nearest Neighbors (k-NN) classifiers.

Significance

The suggested method has many advantages over time-series EEG data analysis and machine learning classifiers used in epilepsy detection studies. A novel image-based preprocessing pipeline using GASF for robust image synthesis and SIFT and ORB for feature extraction is presented here. The study found that the suggested method can accurately discriminate between normal and focal EEG signals, improving patient outcomes through early and accurate epilepsy diagnosis.

A review on the pattern detection methods for epilepsy seizure detection from EEG signals

Over several years, research had been conducted for the detection of epileptic seizures to support an automatic diagnosis system to comfort the clinicians' encumbrance. In this regard, a number of research papers have been published for the identification of epileptic seizures. A thorough review of all these papers is required. So, an attempt has been made to review on the pattern detection methods for epilepsy seizure detection from EEG signals. More than 150 research papers have been discussed to determine the techniques for detecting epileptic seizures. Further, the literature review confirms that the pattern recognition techniques required to detect epileptic seizures varies across the electroencephalogram (EEG) datasets of different conditions. This is mostly owing to the fact that EEG detected under different conditions have different characteristics. This consecutively necessitates the identification of the pattern recognition technique to efficiently differentiate EEG epileptic data from the EEG data of various conditions.

Improved Patient-Independent System for Detection of Electrical Onset of Seizures

PURPOSE

To design a non-patient-specific system to detect the electrical onset of seizures in patients with temporal lobe epilepsy.

METHODS

We used EEG data from 29 seizures of 18 temporal lobe epilepsy patients who underwent multiday video-scalp EEG monitoring as part of their presurgical evaluations. We segmented each data set into preictal and ictal phases, and identified spectral entropy, spectral energy, and signal energy as useful features for discriminating normal and seizure conditions. The performance of five different classifiers was analyzed using these features to design an automated detection system.

RESULTS

Among the five classifiers, decision tree, k-nearest neighbor, and support vector machine performed with sensitivity (specificity) of 79% (81%), 75% (85%), and 80% (86%), respectively. The other two, linear discriminant algorithm and Naive Bayes classifiers, performed with sensitivity (specificity) of 54% (94%), 47% (96%), respectively.

CONCLUSIONS

The support vector machine-based seizure detection system showed better detection capability in terms of sensitivity and specificity measures as compared to linear discriminant algorithm, Naive Bayes, decision tree, and k-nearest neighbor classifiers.

CONCLUSIONS

Our study shows that a generalized system to detect the electrical onset of seizures in temporal lobe epilepsy using scalp-recorded EEG is possible. If confirmed on a larger data set, our findings may have significant implications for the management of seizures, especially in patients with drug-resistant epilepsy.

Epilepsy detection from EEG signals: a review

Over many decades, research is being attempted for the detection of epileptic seizure to support for automatic diagnosis system to help clinicians from burdensome work. In this respect, an enormous number of research papers is published for identification of epileptic seizure. It is difficult to present a detailed review of all these literature. Therefore, in this paper, an attempt has been made to review the detection of an epileptic seizure. More than 100 research papers have been discussed to discern the techniques for detecting the epileptic seizure. Further, the literature survey shows that the pattern recognition required to detect epileptic seizure varies with different conditions of EEG datasets. This is mainly due to the fact that EEG detected under different conditions has different characteristics. This is, in turn, necessitates the identification of pattern recognition technique to effectively distinguish EEG epileptic data from a various condition of EEG data.

Various epileptic seizure detection techniques using biomedical signals: a review

Epilepsy is a chronic chaos of the central nervous system that influences individual’s daily life by putting it at risk due to repeated seizures. Epilepsy affects more than 2% people worldwide of which developing countries are affected worse. A seizure is a transient irregularity in the brain’s electrical activity that produces disturbing physical symptoms such as a lapse in attention and memory, a sensory illusion, etc. Approximately one out of every three patients have frequent seizures, despite treatment with multiple anti-epileptic drugs. According to a survey, population aged 65 or above in European Union is predicted to rise from 16.4% (2004) to 29.9% (2050) and also this tremendous increase in aged population is also predicted for other countries by 2050. In this paper, seizure detection techniques are classified as time, frequency, wavelet (time–frequency), empirical mode decomposition and rational function techniques. The aim of this review paper is to present state-of-the-art methods and ideas that will lead to valid future research direction in the field of seizure detection.

Topolnogical classifier for detecting the emergence of epileptic seizures

OBJECTIVE

An innovative method based on topological data analysis is introduced for classifying EEG recordings of patients affected by epilepsy. We construct a topological space from a collection of EEGs signals using Persistent Homology; then, we analyse the space by Persistent entropy, a global topological feature, in order to classify healthy and epileptic signals.

RESULTS

The performance of the resulting one-feature-based linear topological classifier is tested by analysing the Physionet dataset. The quality of classification is evaluated in terms of the Area Under Curve (AUC) of the receiver operating characteristic curve. It is shown that the linear topological classifier has an AUC equal to [Formula: see text] while the performance of a classifier based on Sample Entropy has an AUC equal to 62.0%.

On the proper selection of preictal period for seizure prediction

Supervised machine learning-based seizure prediction methods consider preictal period as an important prerequisite parameter during training. However, the exact length of the preictal state is unclear and varies from seizure to seizure. We propose a novel statistical approach for proper selection of the preictal period, which can also be considered either as a measure of predictability of a seizure or as the prediction capability of an understudy feature. The optimal preictal periods (OPPs) obtained from the training samples can be used for building a more accurate classifier model. The proposed method uses amplitude distribution histograms of features extracted from electroencephalogram (EEG) recordings. To evaluate this method, we extract spectral power features in different frequency bands from monopolar and space-differential EEG signals of 18 patients suffering from pharmacoresistant epilepsy. Furthermore, comparisons among monopolar channels with space-differential channels, as well as intracranial EEG (iEEG) and surface EEG (sEEG) signals, indicate that while monopolar signals perform better in iEEG recordings, no significant difference is noticeable in sEEG recordings.

Spectral asymmetry and Higuchi's fractal dimension measures of depression electroencephalogram

This study was aimed to compare two electroencephalogram (EEG) analysis methods, spectral asymmetry index (SASI) and Higuchi's fractal dimension (HFD), for detection of depression. Linear SASI method is based on evaluation of the balance of powers in two EEG frequency bands in one channel selected higher and lower than the alpha band spectrum maximum. Nonlinear HFD method calculates fractal dimension directly in the time domain. The resting EEG signals of 17 depressive patients and 17 control subjects were used as a database for calculations. SASI values were positive for depressive and negative for control group (P < 0.05). SASI provided the true detection rate of 88% in the depressive and 82% in the control group. The calculated HFD values detected a small (3%) increase with depression (P < 0.05). HFD provided the true detection rate of 94% in the depressive group and 76% in the control group. The rate of correct indication in the both groups was 85% using SASI or HFD. Statistically significant variations were not revealed between hemispheres (P > 0.05). The results indicated that the linear EEG analysis method SASI and the nonlinear HFD method both demonstrated a good sensitivity for detection of characteristic features of depression in a single-channel EEG.

Spectral asymmetry and Higuchi's fractal dimension measures of depression electroencephalogram

This study was aimed to compare two electroencephalogram (EEG) analysis methods, spectral asymmetry index (SASI) and Higuchi's fractal dimension (HFD), for detection of depression. Linear SASI method is based on evaluation of the balance of powers in two EEG frequency bands in one channel selected higher and lower than the alpha band spectrum maximum. Nonlinear HFD method calculates fractal dimension directly in the time domain. The resting EEG signals of 17 depressive patients and 17 control subjects were used as a database for calculations. SASI values were positive for depressive and negative for control group (P < 0.05). SASI provided the true detection rate of 88% in the depressive and 82% in the control group. The calculated HFD values detected a small (3%) increase with depression (P < 0.05). HFD provided the true detection rate of 94% in the depressive group and 76% in the control group. The rate of correct indication in the both groups was 85% using SASI or HFD. Statistically significant variations were not revealed between hemispheres (P > 0.05). The results indicated that the linear EEG analysis method SASI and the nonlinear HFD method both demonstrated a good sensitivity for detection of characteristic features of depression in a single-channel EEG.

Prediction of seizure likelihood with a long-term, implanted seizure advisory system in patients with drug-resistant epilepsy: a first-in-man study

BACKGROUND

Seizure prediction would be clinically useful in patients with epilepsy and could improve safety, increase independence, and allow acute treatment. We did a multicentre clinical feasibility study to assess the safety and efficacy of a long-term implanted seizure advisory system designed to predict seizure likelihood and quantify seizures in adults with drug-resistant focal seizures.

METHODS

We enrolled patients at three centres in Melbourne, Australia, between March 24, 2010, and June 21, 2011. Eligible patients had between two and 12 disabling partial-onset seizures per month, a lateralised epileptogenic zone, and no history of psychogenic seizures. After devices were surgically implanted, patients entered a data collection phase, during which an algorithm for identification of periods of high, moderate, and low seizure likelihood was established. If the algorithm met performance criteria (ie, sensitivity of high-likelihood warnings greater than 65% and performance better than expected through chance prediction of randomly occurring events), patients then entered an advisory phase and received information about seizure likelihood. The primary endpoint was the number of device-related adverse events at 4 months after implantation. Our secondary endpoints were algorithm performance at the end of the data collection phase, clinical effectiveness (measures of anxiety, depression, seizure severity, and quality of life) 4 months after initiation of the advisory phase, and longer-term adverse events. This trial is registered with ClinicalTrials.gov, number NCT01043406.

FINDINGS

We implanted 15 patients with the advisory system. 11 device-related adverse events were noted within four months of implantation, two of which were serious (device migration, seroma); an additional two serious adverse events occurred during the first year after implantation (device-related infection, device site reaction), but were resolved without further complication. The device met enabling criteria in 11 patients upon completion of the data collection phase, with high likelihood performance estimate sensitivities ranging from 65% to 100%. Three patients' algorithms did not meet performance criteria and one patient required device removal because of an adverse event before sufficient training data were acquired. We detected no significant changes in clinical effectiveness measures between baseline and 4 months after implantation.

INTERPRETATION

This study showed that intracranial electroencephalographic monitoring is feasible in ambulatory patients with drug-resistant epilepsy. If these findings are replicated in larger, longer studies, accurate definition of preictal electrical activity might improve understanding of seizure generation and eventually lead to new management strategies.

FUNDING

NeuroVista.

Effect of camphor essential oil on rat cerebral cortex activity as manifested by fractal dimension changes

The aim of our study was to investigate the effect of camphor essential oil on rat cerebral cortex activity by fractal analysis. Fractal dimension (FD) values of the parietal electrocortical activity were calculated before and after intra-peritoneal administration of camphor essential oil (450-675 ?l/kg) in anesthetized rats. Camphor oil induced seizure-like activity with single and multiple spiking of high amplitudes in the parietal electrocorticogram and occasional clonic limb convulsions. The FD values of cortical activity after camphor oil administration increased on the average. Only FD values of cortical ECoG sequences were lower than those before camphor oil administration.

Progress in predicting seizure episodes with nonlinear methods

One of the most interesting and significant areas of epileptology has been the prediction of the onset of a seizure episode from preictal activity with nonlinear methods. Not only does this type of study have heuristic value for clinical neurophysiology, but it also has potential utilitarian value for the patient with seizures. In this review, 47 reports from 12 centers with multiple studies are presented in chronological order, as are single reports from 21 other centers. The chronological order was chosen to see if progress in the form of earlier prediction was made over time. Only 21% of these reports could provide specific times for the prediction of seizure onset. The range of values was several minutes to 4 hours, with an average (median) of 6-7 minutes. Some reports (16%) had negative or nonspecific findings that prediction times could not be provided. Thus, only limited progress has been made in predicting a seizure from preictal activity, but many other related phenomena have also been studied with nonlinear methods with some success.

Methods of electroencephalographic signal analysis for detection of small hidden changes

The aim of this study was to select and evaluate methods sensitive to reveal small hidden changes in the electroencephalographic (EEG) signal. Two original methods were considered.Multifractal method of scaling analysis of the EEG signal based on the length distribution of low variability periods (LDLVP) was developed and adopted for EEG analysis. The LDLVP method provides a simple route to detecting the multifractal characteristics of a time-series and yields somewhat better temporal resolution than the traditional multifractal analysis.The method of modulation with further integration of energy of the recorded signal was applied for EEG analysis. This method uses integration of differences in energy of the EEG segments with and without stressor.Microwave exposure was used as an external stressor to cause hidden changes in the EEG. Both methods were evaluated on the same EEG database. Database consists of resting EEG recordings of 15 subjects without and with low-level microwave exposure (450 MHz modulated at 40 Hz, power density 0.16 mW/cm2). The significant differences between recordings with and without exposure were detected by the LDLVP method for 4 subjects (26.7%) and energy integration method for 2 subjects (13.3%).The results show that small changes in time variability or energy of the EEG signals hidden in visual inspection can be detected by the LDLVP and integration of differences methods.

A new system theoretic classifier for detection and prediction of epileptic seizures

A system theoretic computational approach has been recently proposed as a generalization of probabilistic networks for modeling complex systems. The computational approach, fuzzy measure-theoretic quantum approximation of an abstract system (FMQAS), generates a system measure between each pair of system objects as a relative measure of association incorporating, through a hierarchical iterative procedure, both the local and global significance of the interaction. FMQAS provides the basis for a new classification algorithm. A preliminary modification of this classification algorithm for temporal sequences is used to analyze electroencephalogram (EEG) data obtained in the temporal neighborhood of a seizure episode to obtain distinct state descriptions (patient invariant characterizations) of the seizure states. This state characterization enables seizure detection before onset with sufficient time to warn the individual or execute actions to abort the seizure formation.

Linear and nonlinear causality between signals: methods, examples and neurophysiological applications

In this paper, we will present and review the most usual methods to detect linear and nonlinear causality between signals: linear Granger causality test (Geweke in J Am Stat Assoc 77:304-313, 1982) extended to direct causality in multivariate case (LGC), directed coherence (DCOH, Saito and Harashima in Recent advances in EEG and EMG data processing, Elsevier, Amsterdam, 1981), partial directed coherence (PDC, Sameshima and Baccala 1999) and nonlinear Granger causality test of Baek and Brock (in Working Paper University of Iowa, 1992) extended to direct causality in multivariate case (partial nonlinear Granger causality, PNGC). All these methods are tested and compared on several ARX, Poisson and nonlinear models, and on neurophysiological data (depth EEG). The results show that LGC, DCOH and PDC are not very robust in relation to nonlinear linkages but they seem to correctly find linear linkages if only the autoregressive parts are nonlinear. PNGC is extremely dependent on the choice of parameters. Moreover, LGC and PNGC may give misleading results in the case of causality on a spectral band, which is illustrated by our neurophysiological database.

Spatio-temporal patient-individual assessment of synchronization changes for epileptic seizure prediction

OBJECTIVE

Abnormal synchronization of neurons plays a central role for the generation of epileptic seizures. Therefore, multivariate time series analysis techniques investigating relationships between the dynamics of different neural populations may offer advantages in predicting epileptic seizures.

METHODS

We applied a phase and a lag synchronization measure to a selected subset of multicontact intracranial EEG recordings and assessed changes in synchronization with respect to seizure prediction.

RESULTS

Patient individual results, group results, spatial aspects using focal and extra-focal electrode contacts as well as two evaluation schemes analyzing decreases and increases in synchronization were examined. Averaged sensitivity values of 60% are observed for a false prediction rate of 0.15 false predictions per hour, a seizure occurrence period of half an hour, and a prediction horizon of 10 min. For approximately half of all 21 patients, a statistically significant prediction performance is observed for at least one synchronization measure and evaluation scheme.

CONCLUSIONS

The results indicate that synchronization changes in the EEG dynamics preceding seizures can be used for seizure prediction. Nevertheless, the underlying pathogenic mechanisms differ and both decreases and increases in synchronization may precede epileptic seizures depending on the structures investigated.

SIGNIFICANCE

The prediction method, optimized values of intervention times, as well as preferred brain structures for the EEG recordings have to be determined for each patient individually offering the chance of a better patient-individual prediction performance.

Correlation dimension and integral do not predict epileptic seizures

Reports in the literature have indicated potential value of the correlation integral and dimension for prediction of epileptic seizures up to several minutes before electrographic onset. We apply these measures to over 2000 total hours of continuous electrocortiogram, taken from 20 patients with epilepsy, examine their sensitivity to quantifiable properties such as the signal amplitude and autocorrelation, and investigate the influence of embedding and filtering strategies on their performance. The results are compared against those obtained from surrogate time series. Our conclusion is that neither the correlation dimension nor the correlation integral has predictive power for seizures.

Non-linear analysis of the electroencephalogram for detecting effects of low-level electromagnetic fields

The study compared traditional spectral analysis and a new scale-invariant method, the analysis of the length distribution of low-variability periods (LDLVPs), to distinguish between electro-encephalogram (EEG) signals with and without a weak stressor, a low-level modulated microwave field. During the experiment, 23 healthy volunteers were exposed to a microwave (450 MHz) of 7 Hz frequency on-off modulation. The field power density at the scalp was 0.16 mW cm(-2). The experimental protocol consisted of ten cycles of repetitive microwave exposure. Signals from frontal EEG channels FP1 and FP2 were analysed. Smooth power spectrum and length distribution curves of low-variability periods, as well as probability distribution close to normal, confirmed that stationarity of the EEG signal during recordings was achieved. The quantitative measure of LDLVPs provided a significant detection of the effect of the stressor for the six subjects exposed to the microwave field but for none of the sham recordings. The spectral analysis revealed a significant result for one subject only. A significant effect of the exposure to the EEG signal was detected in 25% of subjects, with microwave exposure increasing EEG variability. The effect was not detectable by power spectral measures.