Set Of EEG Features Research Articles

Multimodal emotion recognition model via hybrid model with improved feature level fusion on facial and EEG feature set

Multimodal emotion recognition model via hybrid model with improved feature level fusion on facial and EEG feature set

Identifying and differentiating dementias with EEG fractal dimension distributions

AbstractBackgroundAlzheimer’s disease dementia (AD‐dementia) and non‐AD dementia have both been associated with decreases in fractal dimension and spectral slowing in electroencephalography (EEG). Here, we describe a set of EEG features ‐ Fractal Dimension Distributions (FDD) ‐ which quantifies the stability of fractal dimensions over time within specific frequency bands. We compare the performance of models trained using this FDD feature to models using age, spectral power, or overall fractal dimension.MethodWe analyzed resting‐state data from older adults (age M = 71, range = 55‐86) with subjective cognitive impairment (SCI; N = 97, 59 female) or dementia (N = 51, 32 female), including 38 with AD (N = 33). We calculated the Fractal Dimension Distributions (FDD) by first computing the Higuchi fractal dimension within moving windows, then summarizing the distribution of these fractal values using the mean and standard deviation within standard frequency bands. Finally, we trained regularized logistic regression models to classify participants as having SCI or dementia, and as having SCI or AD‐dementia. We compared the performance of models using FDD or standard measures of fractal dimension. Models were evaluated using balanced accuracy, weighted sensitivity, and weighted specificity calculated using 5‐fold cross‐validation.ResultFor SCI vs Dementia, the model using FDD (accuracy = 74%, sensitivity = 67%, specificity = 81%, p < .001) had higher accuracy, sensitivity, and specificity than models trained on age, standard measures of fractal dimension, or spectral power. Similarly, the model trained to predict SCI vs AD‐Dementia was most accurate when using FDD (accuracy = 73%, sensitivity = 59%, specificity = 97%, p < .005). The features useful for discriminating SCI from Dementia partially overlapped with the features useful for discriminating SCI from AD‐Dementia.ConclusionOur novel feature, FDD, outperforms standard measures of fractal dimensions or spectral band power at discriminating between SCI and dementia. Moreover, FDD may help to distinguish between AD and other varieties of dementia.

Epileptic electroencephalography classification using embedded dynamic mode decomposition

Objective. Seizure prediction devices for drug-resistant epileptic patients could lead to improved quality of life and new treatment options, but current approaches to classification of electroencephalography (EEG) segments for early identification of the pre-seizure state typically require many features and complex classifiers. We therefore propose a novel spatio-temporal EEG feature set that significantly aids in separation and easy classification of the interictal and preictal states. Approach. We derive key spectral features from the embedded dynamic mode decomposition (EmDMD) of the brain state system. This method linearizes the complex spatio-temporal dynamics of the system, describing the dynamics in terms of a spectral basis of modes and eigenvalues. The relative subband spectral power and mean phase locking values of these modes prove to be good indicators of the preictal state that precedes seizure onset. Main results. We analyze the linear separability and classification of preictal and interictal states based on our proposed features using seizure data extracted from the CHB-MIT scalp EEG and Kaggle American Epilepsy Society Seizure Prediction Challenge intracranial EEG databases. With a light-weight support vector machine or random forest classifier trained on these features, we classify the preictal state with a sensitivity of up to 92% and specificity of up to 89%. Significance. The EmDMD-derived features separate the preictal and interictal states, improving classification accuracy and motivating further work to incorporate them into seizure prediction algorithms.

Drowsiness detection using portable wireless EEG

Background and ObjectiveThe ever-increasing fatality rate due to traffic and workplace accidents, resulting from drowsiness have been a persistent concern during the past years. An efficient technology capable of monitoring and detecting drowsiness can help to alleviate this concern and has potential applications in driver vigilance monitoring, vigilance monitoring in air traffic control rooms and other safety critical work places. In this paper, we present the feasibility of a wearable light weight wireless consumer grade Electroencephalogram (EEG)-based drowsiness detection. MethodsA set of informative features were extracted from short daytime nap EEG signals and their applicability in discriminating between alert and drowsy state was studied. We derived an optimal set of EEG features, that give maximum detection rate for the drowsy state. In addition, heart rate was also recorded concurrently with EEG and correlation between heart rate and the EEG features corresponding to drowsiness was also studied. ResultsUsing the selected features, the EEG data is shown to be capable of classifying alert and drowsy states with an accuracy of 78.3% using Support Vector Machine classifier employing cross subject validation. The feature selection results also revealed that, the EEG features extracted from the temporal electrodes are more significant for drowsiness detection than the features from frontal electrodes. In addition, EEG features extracted from the temporal electrodes yielded higher correlation coefficient with heart rate, which was in concordance with the feature selection results. ConclusionsThe results reveal that using the proposed drowsiness detection algorithm, it is possible to perform drowsiness detection using a single EEG electrode placed behind the ear.

EEG based dementia diagnosis using multi-class support vector machine with motor speed cognitive test

Dementia is the most burdensome disorder in elders. The Dementia diagnosis is the challenging task at the earliest stages of a neurodegenerative disease when cognitive decline does not interfere with daily life activities. This work focuses on the detection of mild cognitive impairment (MCI) by classifying dementia, MCI and age-matched normal subjects. The classification is based on a different set of EEG features. The multi-class support vector machine (SVM) used to classify EEG features during resting-state, relaxing-state, and motor speed test (MST) events. This work investigated the efficient set of EEG features to calculate maximum classification accuracy for each cognitive event. The motor speed of subjects evaluated and correlated the difference between the dominant and non-dominant hand reactivity with ageing in MST event. The proposed work achieved the highest overall accuracy of 87.59% of MST event after 85.09% in relaxing-state and 80.10% in resting-state. The diagnostic accuracy of MCI group achieved 87.22% in resting-state, 89.72% in relaxing state, and 91.23% in MST. Similarly, dementia achieved 88.72% accuracy in the resting state, 90.23% in relaxing state, and 92.36% in MST event. The normal group achieved 94.66% accuracy in the resting state, 90.23% in relaxing state, and 91.60% in MST event. These findings are comparatively higher than the latest research in this area, and MST findings are novel using multi-class SVM. Thus, MST is the most reliable tool for dementia diagnosis in the clinical setting.

Automated emotion state classification using higher order spectra and interval features of EEG

Automated analysis of electroencephalogram signals for emotion state analysis has been emerging progressively towards the development of affective brain computer interfaces. However, conventional EEG signal analysis techniques such as event related potential (ERP) and power spectrum estimation fail to provide high emotion state classification rates due to Fourier phase suppression. A four-dimensional emotional model in terms of arousal, valence, liking and dominance is proposed using emotion specific EEG signals from DEAP dataset. A compact set of temporal peak/interval related features and spectral features have been extracted to map the feature space. Through the feature map, a multiclass-support vector machine (SVM) based classifier using one-against-one algorithm is configured to yield a maximum classification accuracy of 81.6% while classifying four emotional states. A comparison of multi-class-SVM with other classifiers such as feed forward neural network and radial basis function network has been made. Significant improvement using a proposed compact hybrid EEG feature set and a multi-class-SVM has been achieved for automated emotion state classification.

P2‐364: NEUROANATOMICAL AND NEUROPSYCHOLOGICAL CORRELATES OF RESTING STATE EEG DIAGNOSTIC FEATURES IN PATIENTS WITH ALZHEIMER'S DISEASE

Background: In the search for accurate, low cost biomarkers for Alzheimer’s disease (AD) and other dementias, quantitative electroencephalography (EEG) may offer a solution. In a recent multisite study by Cognision, patients with AD were assessed using the Alzheimer Disease Neuroimaging Initiative protocol, plus EEG assessment. The primary objective of the current analysis, was to examine the relationships between a resting-state (rs)EEG feature set (that best discriminated AD patients from controls) and neuroanatomical measures. The second objective was to identify the rsEEG measures that reflected disease staging. Method: Eighty-nine patients with mild AD (MMSE 21-26) were evaluated using a comprehensive neuropsychological assessment battery, 5 minute eyes-open rsEEG, and structural MRI. Correlations (Spearman’s) were assessed between the 35 rsEEG features (that most accurately discriminated the AD patients), neuroanatomical measures (derived using Freesurfer), and neuropsychological test results. Result: Cortical Thickness (CT) measures within the left posterior cingulate and right precuneus were related to alpha features. Beta features were associated with regions including the right entorhinal cortex, middle temporal, supramarginal, lingual, and paracentral cortex, in addition to the anterior cingulate cortex (ACC) and precuneus, bilaterally. Gamma features correlated with regions that included the right ACC and fronto-parietal cortex. Delta features were linked to the left fronto-parietal and right entorhinal cortex. Theta features were associated with the left ACC and visual cortex. In relation to disease staging – Clinical Dementia Rating scores were correlated with gamma features at frontal electrode sites, and with power over frequency bands, delta to beta, at Fz. Alpha features were associated with hippocampal volume (bilaterally), whereas some delta and theta features were linked to left hippocampal volume. Conclusion: These preliminary correlation analyses highlight multiple brain regions that appear to underpin the rsEEG abnormalities that occur due to AD. Given the rich data offered by both rsEEG and by structural MRI, future studies could investigate the combined potential for these techniques to classify the dementias.

Optimal set of EEG features in infant sleep stage classification

This paper evaluates six classification algorithms to assess the importance of individual EEG rhythms in the context of automatic classification of infant sleep. EEG features were obtained by Fourier transform and by a novel technique based on the empirical mode decomposition and generalized zero crossing method. Of six evaluated classification algorithms, the best classification results were obtained with the support vector machine for the combination of all presented features from four EEG channels. Three methods of attribute ranking were assessed: relief, principal component analysis, and wrapper-based optimized attribute weights. The outcomes revealed that the optimal selection of features requires one feature from every significant frequency band, either a spectral feature or a frequency dynamic feature. This means that reducing the number of features will have a minimal impact on the classification accuracy.

EEG may serve as a biomarker in Huntington\u2019s disease using machine learning automatic classification

Reliable markers measuring disease progression in Huntington’s disease (HD), before and after disease manifestation, may guide a therapy aimed at slowing or halting disease progression. Quantitative electroencephalography (qEEG) may provide a quantification method for possible (sub)cortical dysfunction occurring prior to or concomitant with motor or cognitive disturbances observed in HD. In this pilot study we construct an automatic classifier distinguishing healthy controls from HD gene carriers using qEEG and derive qEEG features that correlate with clinical markers known to change with disease progression in HD, with the aim of exploring biomarker potential. We included twenty-six HD gene carriers (49.7 ± 8.5 years) and 25 healthy controls (52.7 ± 8.7 years). EEG was recorded for three minutes with subjects at rest. An EEG index was created by applying statistical pattern recognition to a large set of EEG features, which was subsequently tested using 10-fold cross-validation. The index resulted in a continuous variable ranging from 0 to 1: a low value indicating a state close to normal and a high value pointing to HD. qEEG features that correlate specifically with commonly used clinical markers in HD research were derived. The classification index had a specificity of 83%, a sensitivity of 83% and an accuracy of 83%. The area under the curve of the receiver operator characteristic curve was 0.9. qEEG analysis on subsets of electrophysiological features resulted in two highly significant correlations with clinical scores. The results of this pilot study suggest that qEEG may serve as a biomarker in HD. The indices correlating with modalities changing with the progression of the disease may lead to tools based on qEEG that help monitor efficacy in intervention studies.

EEG analysis of seizure patterns using visibility graphs for detection of generalized seizures

BackgroundThe traditional EEG features in the time and frequency domain show limited seizure detection performance in the epileptic population with intellectual disability (ID). In addition, the influence of EEG seizure patterns on detection performance was less studied. New methodA single-channel EEG signal can be mapped into visibility graphs (VGS), including basic visibility graph (VG), horizontal VG (HVG), and difference VG (DVG). These graphs were used to characterize different EEG seizure patterns. To demonstrate its effectiveness in identifying EEG seizure patterns and detecting generalized seizures, EEG recordings of 615h on one EEG channel from 29 epileptic patients with ID were analyzed. ResultsA novel feature set with discriminative power for seizure detection was obtained by using the VGS method. The degree distributions (DDs) of DVG can clearly distinguish EEG of each seizure pattern. The degree entropy and power-law degree power in DVG were proposed here for the first time, and they show significant difference between seizure and non-seizure EEG. The connecting structure measured by HVG can better distinguish seizure EEG from background than those by VG and DVG. Comparison with existing methodA traditional EEG feature set based on frequency analysis was used here as a benchmark feature set. With a support vector machine (SVM) classifier, the seizure detection performance of the benchmark feature set (sensitivity of 24%, FDt/h of 1.8s) can be improved by combining our proposed VGS features extracted from one EEG channel (sensitivity of 38%, FDt/h of 1.4s). ConclusionsThe proposed VGS-based features can help improve seizure detection for ID patients.

Optimal set of EEG features for emotional state classification and trajectory visualization in Parkinson's disease

In addition to classic motor signs and symptoms, individuals with Parkinson's disease (PD) are characterized by emotional deficits. Ongoing brain activity can be recorded by electroencephalograph (EEG) to discover the links between emotional states and brain activity. This study utilized machine-learning algorithms to categorize emotional states in PD patients compared with healthy controls (HC) using EEG. Twenty non-demented PD patients and 20 healthy age-, gender-, and education level-matched controls viewed happiness, sadness, fear, anger, surprise, and disgust emotional stimuli while fourteen-channel EEG was being recorded. Multimodal stimulus (combination of audio and visual) was used to evoke the emotions. To classify the EEG-based emotional states and visualize the changes of emotional states over time, this paper compares four kinds of EEG features for emotional state classification and proposes an approach to track the trajectory of emotion changes with manifold learning. From the experimental results using our EEG data set, we found that (a) bispectrum feature is superior to other three kinds of features, namely power spectrum, wavelet packet and nonlinear dynamical analysis; (b) higher frequency bands (alpha, beta and gamma) play a more important role in emotion activities than lower frequency bands (delta and theta) in both groups and; (c) the trajectory of emotion changes can be visualized by reducing subject-independent features with manifold learning. This provides a promising way of implementing visualization of patient's emotional state in real time and leads to a practical system for noninvasive assessment of the emotional impairments associated with neurological disorders.

A Framework for Content-based Retrieval of EEG with Applications to Neuroscience and Beyond.

This paper introduces a prototype framework for content-based EEG retrieval (CBER). Like content-based image retrieval, the proposed framework retrieves EEG segments similar to the query EEG segment in a large database. Such retrieval of EEG can be used to assist data mining of brain signals by allowing researchers to understand the association between brain patterns, responses, and the environment. Retrieval might also be used to enhance the accuracy of Brain Computer Interface (BCI) systems by providing related samples for training. We present key components of CBER and explain how to handle the distinctive characteristics of EEG. To demonstrate the feasibility of the framework, we implemented a simple EEG database of about 37,000 samples from more than 100 subjects. We ran two retrieval scenarios with a set of EEG features and evaluation metrics. The results of finding similar subjects clearly demonstrate the potential of CBER in many EEG applications.

EEG study of the neural representation and classification of semantic categories of animals vs tools in young and elderly participants

Our semantic knowledge is essential for unimpaired cognition, language and behavior. Previously, we have used Electroencephalography (EEG) to decode the semantic categories of animals vs tools in younger adult subjects during a covert image naming task[1]. In this work, we are extending this approach to older subjects. To date we have recorded EEG from 6 normal older controls. A set of EEG features were extracted from the time-domain at a range of temporal scales. These signal amplitude features were used as inputs to L2-penalised logistic regression classifier, after univariate Anova feature selection. Figure ​Figure11 compares the classification results of the older group (60-79yrs, mean 68) to the previous cohort of 7 younger participants (25-33yrs, mean 29). The mean accuracy for these two groups increases similarly with the number of channels used as input to the classifier, and reaches a saturation value around 75% (Figure ​(Figure1A),1A), where baseline is 50%, and an accuracy >56% is significant at α = 0.05. Using a narrow sliding window to extract features (width of 25 ms, moving in steps of 10 ms) over the period 200 ms before to 1000 ms after stimulus onset, we found that the accuracy peaks higher and earlier in the younger subjects (Figure ​(Figure1B),1B), but the contrast only approached significance with the current sample size. We also calculated the mean classification accuracy of each channel and plotted them as scalp-maps. The results show that the accuracy is higher in right frontal and occipital regions in younger subjects (Figure ​(Figure1C),1C), while the accuracy is higher in occipital, parietal and left frontal areas in older subjects (Figure ​(Figure1D1D). Figure 1 Comparison of classification accuracy of animals vs tools between younger and older groups.

Reliable epileptic seizure detection using an improved wavelet neural network

Electroencephalogram (EEG) signal analysis is indispensable in epilepsy diagnosis as it offers valuable insights for locating the abnormal distortions in the brain wave. However, visual interpretation of the massive amounts of EEG signals is time-consuming, and there is often inconsistent judgment between experts. This study proposes a novel and reliable seizure detection system, where the statistical features extracted from the discrete wavelet transform are used in conjunction with an improved wavelet neural network (WNN) to identify the occurrence of seizures. Experimental simulations were carried out on a well-known publicly available dataset, which was kindly provided by the Epilepsy Center, University of Bonn, Germany. The normal and epileptic EEG signals were first pre-processed using the discrete wavelet transform. Subsequently, a set of statistical features was extracted to train a WNNs-based classifier. The study has two key findings. First, simulation results showed that the proposed improved WNNs-based classifier gave excellent predictive ability, where an overall classification accuracy of 98.87% was obtained. Second, by using the 10th and 90th percentiles of the absolute values of the wavelet coefficients, a better set of EEG features can be identified from the data, as the outliers are removed before any further downstream analysis. The obtained high prediction accuracy demonstrated the feasibility of the proposed seizure detection scheme. It suggested the prospective implementation of the proposed method in developing a real time automated epileptic diagnostic system with fast and accurate response that could assist neurologists in the decision making process.

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.