Noisy EEG Signals Research Articles

Reading into the mind’s eye: Boosting automatic visual recognition with EEG signals

Classifying visual information is an apparently simple and effortless task in our everyday routine, but can we automatically predict what we see from signals emitted by the brain?While other researchers have already attempted to answer this question, we are the first to show that a commercially available BCI could be effectively used for visual image classification in real-world scenarios – when testing takes place at a completely different time than training data collection. The task is difficult, as it requires relating the noisy and low-level EEG signals to complex and highly semantic visual categories. In this paper, we propose different learning approaches and show that simpler classifiers such as Ridge Regression with Gabor filtering of the input EEG signal could be more effective than the powerful Long Short Term Memory Networks and Convolutional Neural Networks in this case of limited and noisy training data. We analyzed the importance of each electrode for the visual classification task and noticed that the sensors with the highest accuracy were the ones that recorded brain activity from regions known to be correlated more with higher level recognition and cognitive processes and less to lower-level visual signal processing. The result is also in accordance with research in computer vision with deep neural networks, which shows that semantic visual features are learned only at higher levels of neural depth.While EEG signals are weaker by themselves for the task of visual classification, we demonstrate that they could be powerful when combined with deep visual features extracted from the image, improving performance from 91% to over 97% in a multi-class recognition setting. Our tests show that EEG input brings additional information that is not learned by artificial deep networks on the given image training set. Thus, a commercially available BCI could be effectively used in conjunction with a deep learning based vision system to form together a stronger visual recognition system that is suitable for real-world applications.

An Efficient Mixture Model Approach in Brain-Machine Interface Systems for Extracting the Psychological Status of Mentally Impaired Persons Using EEG Signals

We propose an efficient mixture classification technique, which uses electroencephalography (EEG) signals for establishing a communication channel for the physically challenged or immobilized people, by the usage of the brain signals. In order to identify the emotion expressions by an immobilized person, we introduce a novel approach for emotion recognition based on the generalized mixture distribution model. The main benefit of utilizing this model is that it is an asymmetric distribution, which helps to extract the EEG signals, which are either in symmetric or asymmetric form. The skew Gaussian distribution helps to identify the small duration EEG signal sample and helps toward better recognition of emotions in both clean and noisy EEG signals. The proposed method is particularly well suited for the high variability of the EEG signal allowing the emotions to be identified appropriately. The features of the brain signals are extracted by using cepstral coefficients. The extracted features are classified into different emotions using mixture classification techniques. In order to validate the model, six mentally impaired subjects are considered in the age group of 60-68, and an 8-channel EEG signal is utilized to collect the EEG signals under audio-visual stimuli. The basic emotions considered in this study include happy, sad, neutral, and boredom and an average emotion recognition accuracy of 89% is achieved.

Developing a Multi-channel Beamformer by Enhancing Spatially Constrained ICA for Recovery of Correlated EEG Sources.

Background: Brain source imaging based on electroencephalogram (EEG) data aims to recover the neuron populations’ activity producing the scalp potentials. This procedure is known as the EEG inverse problem. Recently, beamformers have gained a lot of consideration in the EEG inverse problem. Objective: Beamformers lack acceptable performance in the case of correlated brain sources. These sources happen when some regions of the brain have simultaneous or correlated activities such as auditory stimulation or moving left and right extremities of the body at the same time. In this paper, we have developed a multichannel beamformer robust to correlated sources. Material and Methods: In this simulation study, we have looked at the problem of brain source imaging and beamforming from a blind source separation point of view. We focused on the spatially constraint independent component analysis (scICA) algorithm, which generally benefits from the pre-known partial information of mixing matrix, and modified the steps of the algorithm in a way that makes it more robust to correlated sources. We called the modified scICA algorithm Multichannel ICA based EEG Beamformer (MIEB). Results: We evaluated the proposed algorithm on simulated EEG data and compared its performance quantitatively with three algorithms scICA, linearly-constrained minimum-variance (LCMV) and Dual-Core beamformers; it is considered that the latter is specially designed to reconstruct correlated sources. Conclusion: The MIEB algorithm has much better performance in terms of normalized mean squared error in recovering the correlated/uncorrelated sources both in noise free and noisy synthetic EEG signals. Therefore, it could be used as a robust beamformer in recovering correlated brain sources.

Using a multichannel Wiener filter to remove eye-blink artifacts from EEG data

Abstract This paper presents a novel method for removing ocular artifacts from EEG recordings. The proposed approach is based on time-domain linear filtering. Instead of directly estimating the artifact-free signal, we propose to obtain the eye-blink signal first, using a multichannel Wiener filter (MWF) and a small subset of the frontal electrodes, so that extra EOG sensors are unnecessary. Then, the estimate of the eye-blink signal is subtracted from the noisy EEG signal in accordance with principles of regression analysis. We have performed numerical simulations so as to compare our approach to the independent component analysis (ICA) that is commonly used in EEG enhancement. Our experiments show that the MWF-based approach can perform better than the ICA in terms of eye-blink cancellation and signal distortions. Besides that, the proposed approach is conceptually simpler and better suited to real-time applications.

Embedding Dimension Selection for Adaptive Singular Spectrum Analysis of EEG Signal.

The recorded electroencephalography (EEG) signal is often contaminated with different kinds of artifacts and noise. Singular spectrum analysis (SSA) is a powerful tool for extracting the brain rhythm from a noisy EEG signal. By analyzing the frequency characteristics of the reconstructed component (RC) and the change rate in the trace of the Toeplitz matrix, it is demonstrated that the embedding dimension is related to the frequency bandwidth of each reconstructed component, in consistence with the component mixing in the singular value decomposition step. A method for selecting the embedding dimension is thereby proposed and verified by simulated EEG signal based on the Markov Process Amplitude (MPA) EEG Model. Real EEG signal is also collected from the experimental subjects under both eyes-open and eyes-closed conditions. The experimental results show that based on the embedding dimension selection method, the alpha rhythm can be extracted from the real EEG signal by the adaptive SSA, which can be effectively utilized to distinguish between the eyes-open and eyes-closed states.

Performance of an automated algorithm to process artefacts for quantitative EEG analysis during a simultaneous driving simulator performance task

BackgroundArtefact removal from noisy EEG signals is cumbersome, and often requires manual intervention. We tested the performance of an automated method to detect and remove artefacts from EEG recorded during a driving simulation task. MethodsFive patients with obstructive sleep apnea (OSA) and five healthy controls were randomly selected from 17 participants undergoing a 40-h extended wakefulness study with 2-hourly 30-minute simulated driving tasks with simultaneous EEG. Two EEG recordings from each individual were studied. EEG data was first processed by independent component analysis (ICA). The accuracy of the automated algorithm (AA) to detect residual EEG artefact was evaluated against a reference-standard (RS) of visually identified artefact-contaminated epochs. EEG spectral power was calculated using 1) the RS method, 2) the AA, and 3) raw data without any artefact rejection (ICA only). ResultsThe algorithm showed good sensitivity (median: 83.9%), excellent specificity (91.1%), and high accuracy (87.0%) to detect noisy epochs. Cohen's κ indicated a substantial agreement between the two methods (0.72). EEG spectral power calculated using the RS and the AA did not differ significantly, while the power of the raw signal was significantly higher than those produced by any artefact rejection method. Increased EEG delta and theta power were significantly correlated with poorer driving performance. ConclusionsThese preliminary findings demonstrate an effective automated method to process EEG artefact recorded during driving simulation. This approach may facilitate the routine application of quantitative EEG analyses in future studies and identify new markers of impaired driving performance associated with sleep disorders.

Investigation of the Effect of Video Game Play on Cognitive Behaviour using Adaptive Noise Cancellation and Wavelet Transform

This work shows the effect on cognitive performance while playing action video game based on the beta waves variation. The Present work initially shows the Signal to Noise Ratio (SNR) improvement by using an adaptive noise cancellation technique for suppressing the noise, here is to eliminate the ECG artifact from noisy EEG signal and then the wavelet analysis of EEG signals is done to study the brain activity. EEG signal is acquired experimentally from the subject by using the 10-20 system of electrode placement. In first stage the adaptive noise cancellation technique is implemented with the help of an adaptive filter to remove the artifact and hence to improve the signal to noise ratio (SNR). The adaptive algorithm used for the filter is the LMS which is a FIR based digital filter. A synthetic signal is also simulated to give a concept of the effect of the proposed active noise cancellation method to remove the noise or artifact. Synthetic signal analysis shows the suitability of the adaptive noise cancellation technique which is a de-noising tool, and here the de-noised signal is the error signal. Then in second stage, the de-noised signal is processed and the Wavelet Transform is used to provide the information of the β-activity. EEG signals from the frontal and parietal lobes are recorded and the effect is analyzed as the frontal and parietal lobe is associated with reasoning, planning, problem solving and orientation, recognition and perception of stimuli respectively. However, the presence of artifacts like electrocardiogram(ECG) in the EEG signal is a major problem in the study of brain activity. Though it is needed to record cerebral signals, it also records the signals that are not of cerebral origin called artifacts. Artifact removal from EEG signals is essential for better analysis and diagnosis. The EEG signal acquired from a patient is generally noisy and non-stationary by nature. So an adaptive noise cancellation technique is used to remove the artifact and hence the signal to noise ratio of the acquired EEG signal is improved. The EEG signal alone, is acquired from the subject and is used as the reference signal in the ANC noise cancellation technique to get the de-noise EEG signal. Then the de-noised signal is processed and the Wavelet Transform (WT) is used to provide the information of the β-activity. Experiments reveal that the proposed ANC method of noise or artifact removal along with the Wavelet Transform can be used to provide a clear information of the β-activity during the action video game play. The increase in the Beta activity in the eleventh hour of game play than the game play of the first hour indicates increase in cognition due to training in action video game play.

EEG signal enhancement using cascaded S-Golay filter

Electroencephalogram (EEG) is the most popular signal used for diagnosis of brain disorders. A good quality EEG signal provides the proper interpretation and identification of physiological and pathological phenomena. However, these recordings are often corrupted by different kinds of noise. As Savitzky Golay smoothing filter (SGSF) preserves the peaks and minimize the signal distortion, its use in cascade may further enhance this capability. Therefore in the present work cascaded SGSF (CSGSF) is proposed to filter the noisy EEG signals. The CSGSF combines two successive Savitzky Golay filters. For comparative analysis, other cascaded arrangements like cascaded moving average filter (CMAF), MAF-SGSF, SGSF-Binomial and single stage SGSF are also designed. These filters are tested on artificial EEG signals added with white Gaussian noise and non Gaussian noise. These filters are also tested on real time EEG signals. The filtered signals are assessed through signal to noise ratio (SNR), signal to signal plus noise ratio (SSNR), SNR improvement (SNRI), mean square error (MSE) and correlation coefficient (COR). It is revealed from the results that CSGSF outperforms the other designed filters in case of artificial and real time EEG signals.

Algorithm based on the short-term Rényi entropy and IF estimation for noisy EEG signals analysis

Stochastic electroencephalogram (EEG) signals are known to be nonstationary and often multicomponential. Detecting and extracting their components may help clinicians to localize brain neurological dysfunctionalities for patients with motor control disorders due to the fact that movement-related cortical activities are reflected in spectral EEG changes. A new algorithm for EEG signal components detection from its time-frequency distribution (TFD) has been proposed in this paper. The algorithm utilizes the modification of the Rényi entropy-based technique for number of components estimation, called short-term Rényi entropy (STRE), and upgraded by an iterative algorithm which was shown to enhance existing approaches. Combined with instantaneous frequency (IF) estimation, the proposed method was applied to EEG signal analysis both in noise-free and noisy environments for limb movements EEG signals, and was shown to be an efficient technique providing spectral description of brain activities at each electrode location up to moderate additive noise levels. Furthermore, the obtained information concerning the number of EEG signal components and their IFs show potentials to enhance diagnostics and treatment of neurological disorders for patients with motor control illnesses.

Evaluating Quantum Neural Network filtered motor imagery brain-computer interface using multiple classification techniques

The raw EEG signal acquired non-invasively from the sensorimotor cortex during the motor imagery (MI) performed by a brain-computer interface (BCI) user is naturally embedded with noise while the actual noise-free EEG is still unattainable. This paper compares the enhancement in information when filtering these noisy EEG signals while using a Schrodinger wave equation (SWE) based Recurrent Quantum Neural Network (RQNN) model and a Savitzky–Golay (SG) filtering model, while investigating over multiple classification techniques on several datasets. The RQNN model is designed to efficiently capture the statistical behavior of the input signal using an unsupervised learning scheme. The algorithm is robust to parametric sensitivity and does not make any a priori assumption about the true signal type or the embedded noise. The performance of both the filtering approaches, investigated for the BCI competition IV 2b dataset as well as the offline datasets on subjects in the BCI laboratory, over multiple classifiers shows that the RQNN can potentially be a flexible technique that can suit different classifiers for real-time EEG signal filtering. The average classification accuracy performance across all the subjects with the RQNN technique is better than the SG (and the unfiltered signal) by approximately 5% (and 7%) and 1% (and 4%) during the training and the evaluation stages respectively.

A Beamformer-Particle Filter Framework for Localization of Correlated EEG Sources.

Electroencephalography (EEG)-based brain computer interface (BCI) is the most studied noninvasive interface to build a direct communication pathway between the brain and an external device. However, correlated noises in EEG measurements still constitute a significant challenge. Alternatively, building BCIs based on filtered brain activity source signals instead of using their surface projections, obtained from the noisy EEG signals, is a promising and not well-explored direction. In this context, finding the locations and waveforms of inner brain sources represents a crucial task for advancing source-based noninvasive BCI technologies. In this paper, we propose a novel multicore beamformer particle filter (multicore BPF) to estimate the EEG brain source spatial locations and their corresponding waveforms. In contrast to conventional (single-core) beamforming spatial filters, the developed multicore BPF considers explicitly temporal correlation among the estimated brain sources by suppressing activation from regions with interfering coherent sources. The hybrid multicore BPF brings together the advantages of both deterministic and Bayesian inverse problem algorithms in order to improve the estimation accuracy. It solves the brain activity localization problem without prior information about approximate areas of source locations. Moreover, the multicore BPF reduces the dimensionality of the problem to half compared with the PF solution, thus alleviating the curse of dimensionality problem. The results, based on generated and real EEG data, show that the proposed framework recovers correctly the dominant sources of brain activity.

Ensemble classifier for epileptic seizure detection for imperfect EEG data.

Brain status information is captured by physiological electroencephalogram (EEG) signals, which are extensively used to study different brain activities. This study investigates the use of a new ensemble classifier to detect an epileptic seizure from compressed and noisy EEG signals. This noise-aware signal combination (NSC) ensemble classifier combines four classification models based on their individual performance. The main objective of the proposed classifier is to enhance the classification accuracy in the presence of noisy and incomplete information while preserving a reasonable amount of complexity. The experimental results show the effectiveness of the NSC technique, which yields higher accuracies of 90% for noiseless data compared with 85%, 85.9%, and 89.5% in other experiments. The accuracy for the proposed method is 80% when SNR = 1 dB, 84% when SNR = 5 dB, and 88% when SNR = 10 dB, while the compression ratio (CR) is 85.35% for all of the datasets mentioned.

EEG data analysis based on EMD for coma and quasi-brain-death patients

Electroencephalography (EEG) is widely used in evaluating the absence of cerebral cortex function for the determination of brain death. Since EEG recorded signal is always corrupted by some artefacts and various interfering noise, extracting active or nonactive features from noisy EEG signals and evaluating their significance is therefore crucial in the process of brain death diagnosis. This article presents an EEG-based preliminary examination system associated with empirical mode decomposition (EMD) technique to extract informative brain activity features from real-world recorded clinical EEG data. Moreover, the power spectrum technique is applied to evaluate the significant differences between the group of comatose patients and the group of quasi-brain-deaths. Our experimental results show effectiveness and some promising directions of applying the EMD method to the clinical EEG analysis.

A Novel Method for Detection of Epilepsy in Short and Noisy EEG Signals Using Ordinal Pattern Analysis

Introduction: In this paper, a novel complexity measure is proposed to detect dynamical changes in nonlinear systems using ordinal pattern analysis of time series data taken from the system. Epilepsy is considered as a dynamical change in nonlinear and complex brain system. The ability of the proposed measure for characterizing the normal and epileptic EEG signals when the signal is short or is contaminated with noise is investigated and compared with some traditional chaos-based measures. Materials and Methods: In the proposed method, the phase space of the time series is reconstructed and then partitioned using ordinal patterns. The partitions can be labeled using a set of symbols. Therefore, the state trajectory is converted to a symbol sequence. A finite state machine is then constructed to model the sequence. A new complexity measure is proposed to detect dynamical changes using the state transition matrix of the state machine. The proposed complexity measure was applied to detect epilepsy in short and noisy EEG signals and the results were compared with some chaotic measures. Results: The results indicate that this complexity measure can distinguish normal and epileptic EEG signals with an accuracy of more than 97% for clean EEG and more than 75% for highly noised EEG signals. Discussion and Conclusion: The complexity measure can be computed in a very fast and easy way and, unlike traditional chaotic measures, is robust with respect to noise corrupting the data. This measure is also capable of dynamical change detection in short time series data.

A self-paced brain–computer interface system with a low false positive rate

The performance of current EEG-based self-paced brain–computer interface (SBCI) systems is not suitable for most practical applications. In this paper, an improved SBCI that uses features extracted from three neurological phenomena (movement-related potentials, changes in the power of Mu rhythms and changes in the power of Beta rhythms) to detect an intentional control command in noisy EEG signals is proposed. The proposed system achieves a high true positive (TP) to false positive (FP) ratio. To extract features for each neurological phenomenon in every EEG signal, a method that consists of a stationary wavelet transform followed by matched filtering is developed. For each neurological phenomenon in every EEG channel, features are classified using a support vector machine classifier (SVM). For each neurological phenomenon, a multiple classifier system (MCS) then combines the outputs of the SVMs. Another MCS combines the outputs of MCSs designed for the three neurological phenomena. Various configurations for combining the outputs of these MCSs are considered. A hybrid genetic algorithm (HGA) is proposed to simultaneously select the features, the values of the classifiers' parameters and the configuration for combining MCSs that yield the near optimal performance. Analysis of the data recorded from four able-bodied subjects shows a significant performance improvement over previous SBCIs.

Approximate Time-Variable Coherence Analysis of Multichannel Signals

We present a new method for signal extraction from noisy multichannel epileptic seizure onset EEG signals. These signals are non-stationary which makes time-invariant filtering unsuitable. The new method assumes a signal model and performs denoising by filtering the signal of each channel using a time-variable filter which is an estimate of the Wiener filter. The approximate Wiener filters are obtained using the time-frequency coherence functions between all channel pairs, and a fix-point algorithm. We estimate the coherence functions using the multiple window method, after which the fix-point algorithm is applied. Simulations indicate that this method improves upon its restriction to assumed stationary signals for realistically non-stationary data, in terms of mean square error, and we show that it can also be used for time-frequency representation of noisy multichannel signals. The method was applied to two epileptic seizure onset signals, and it turned out that the most informative output of the method are the filters themselves studied in the time-frequency domain. They seem to reveal hidden features of the epileptic signal which are otherwise invisible. This algorithm can be used as preprocessing for seizure onset EEG signals prior to time-frequency representation and manual or algorithmic pattern classification.

Development of Newton-type adaptive algorithm for minimization of EOG artefacts from noisy EEG signals

In this paper, we propose an adaptive algorithm using Newton's method for the enhancement of EEG signals in the presence of EOG artefacts. We consider two models for the noise (i.e., artefacts) estimation: (i) the conventional linear model and (ii) a nonlinear model using second-order Volterra function. Application of Newton's method to the linear case results in the conventional recursive least-squares algorithm. Since the parameters of the nonlinear model are specified in terms of a vector and a matrix, the conventional Newton's method could not be applied. Hence, the underlying cost function has been reformulated whereby all the parameters are represented by a single vector and then Newton's method is applied to this reformulated cost function. While developing the algorithm for the nonlinear case, the Hessian matrix is approximated because of the special property of the reference signal. This ensures the reduced computational complexity and positive definiteness of the approximated Hessian so as to ensure search along the descent directions. Another algorithm making use of the exact Hessian matrix is also derived. These algorithms were used to minimize the EOG artefacts from EEG signals. Simulation results show that the nonlinear scheme with approximated Hessian works well compared to the other two algorithms in minimizing EOG artefacts from contaminated EEG signals.