Comparative Analysis of Spectral Approaches to Feature Extraction for EEG-Based Motor Imagery Classification (original) (raw)
Related papers
American Journal of Biomedical Engineering, 2013
Brain Computer Interface (BCI) has attracted many researchers in recent years, in which one of non-invasive techniques for the BCI issue is Electroencephalography (EEG). This EEG technique can allow to investigate human brain related to muscle for diagnosis and rehabilitation. In this paper, we proposed an Average Partial Power Spectrum Density (APPSD) approach and a bandpass filter to classify mental tasks using an EGG system with 24 channels, in which the relevant mental tasks are: left hand movement, right hand movement and rest. The proposed approach is the combination of the 2 Hz bandpass filter and the APPSD algorithm in the specific frequency ranges to find out features of imagery for classification. For the accuracy of the feature extraction, outliers which sparsely appear in the range of the PSD are removed. From the obtained features of movements, an Artificial Neuron Network (ANN) model was used to classify imagery status. Experiments were performed on 2 subjects with 200 runs per one subject to illustrate the effectiveness of the proposed method.
Motor imagery (MI) refers to the mental representation of movement without any motor action. Effective classification of MI tasks is promising for patients with motor disabilities. However, achieving a reliable MI task classification by electroencephalography (EEG) is challenging. The purpose of the study is to improve classification performance by providing discriminative features of intent in MI tasks. In this study, a matrix decomposition technique, namely Non-Negative Matrix Factorization (NMF) is adopted for feature extraction from time-frequency distributions (TFD) obtained by the Waveletbased Synchrosqueezing Transform (WSST). EEG signals were meticulously cleared of noise by Independent Component Analysis (ICA) before performing NMF. Then, the various features such as skewness, kurtosis, discontinuity, standard deviation and sparsity were extracted from NMF vectors in order to train five different classifiers, namely Support Vector Machine (SVM), K-Nearest Neighbor (KNN), Decision Tree (DT), Random Forest (RF) and AdaBoost. WSST provides energy distributions with highly localization capability in TFD, using the reassignment technique. TFDs showed distinctive energy spikes for rightand left-hand MI tasks. The experimental results revealed that the proposed method approaches outstanding accuracy (>95%), kappa (>0.90), and F1 score (0.99) by using KNN, RF, DT and AdaBoost. The study results also show that WSST-based NMF features are promising for MI task classification.
Conference proceedings : ... Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Annual Conference, 2006
We introduce an adaptive space time frequency analysis to extract and classify subject specific brain oscillations induced by motor imagery in a brain computer interface task. The introduced method requires no prior knowledge of the reactive frequency bands, their temporal behavior or cortical locations. The algorithm implements an arbitrary time-frequency segmentation procedure by using a flexible local discriminant base algorithm for given multichannel brain activity recordings to extract subject specific ERD and ERS patterns. Extracted time-frequency features are processed by principal component analysis to reduce the feature set which is highly correlated due to volume conduction and the neighbor cortical regions. The reduced feature set is then fed to a linear discriminant analysis for classification. We give experimental results for 9 subjects to show the superior performance of the proposed method where the classification accuracy varied between 76.4% and 96.8% and the averag...
2014 5th International Conference on Intelligent Systems, Modelling and Simulation, 2014
Brain computer interface (BCI) provides an interface between a brain and a computer in order to enable people to control external devices without using muscles. In this work, authors report on the results of implementation of three algorithms using wavelet features collected with different kinds of features during imagining left hand, right hand, and foot movements. The features of event-related desynchronization (ERD/ERS) were extracted from alpha and beta frequency bands, and followed by one classifier among the three following ones; linear discriminant analysis (LDA), support vector machine (SVM) or K-nearest neighbor (KNN). The data were recorded from three subjects, provided by BCI-Competition III. The performance evaluation of the proposed algorithms was provided by Matab simulation. The best combination was the wavelet coefficients and common spatial pattern algorithms, followed by the suppor vector machine classifier with an average classification accuracy of 75%, which is an interesting for motor imagery application.
Computer Methods and Programs in Biomedicine, 2020
Background and Objective: Motor Imagery (MI) based Brain-Computer-Interface (BCI) is a rising support system that can assist disabled people to communicate with the real world, without any external help. It serves as an alternative communication channel between the user and computer. Electroencephalogram (EEG) recordings prove to be an appropriate choice for imaging MI tasks in a BCI system as it provides a non-invasive way for completing the task. The reliability of a BCI system confides on the efficiency of the assessment of different MI tasks. Methods: The present work proposes a new approach for the classification of distinct MI tasks based on EEG signals using the flexible analytic wavelet transform (FAWT) technique. The FAWT decomposes the EEG signal into sub-bands and temporal moment-based features are extracted from the sub-bands. Feature normalization is applied to minimize the bias nature of classifier. The FAWT-based features are utilized as inputs to multiple classifiers. Ensemble learning method based Subspace k-Nearest Neighbour (kNN) classifier is established as the best and robust classifier for the distinction of the right hand (RH) and right foot (RF) MI tasks. Results: The sub-band (SB) wise features are tested on multiple classifiers and best performance parameters are obtained using the ensemble method based subspace kNN classifier. The best results of parameters are obtained for fourth SB as accuracy 99.33%, sensitivity 99%, specificity 99.6%, F1-Score 0.9925, and kappa value 0.9865. The other sub-bands are also attained significant results using subspace KNN classifier. Conclusions: The proposed work explores the utility of FAWT based features for the classification of RH and RF MI tasks EEG signals. The suggested work highlights the effectiveness of multiple classifiers for classification MI-tasks. The proposed method shows better performance in comparison to state-of-arts methods. Thus, the potential to implement a BCI system for controlling wheelchairs, robotic arms, etc.
Multi-Sessions Outcome for EEG Feature Extraction and Classification Methods in a Motor Imagery Task
Traitement du Signal, 2021
Received: 21 August 2020 Accepted: 20 March 2021 The purpose of this research is to evaluate the performances of some features extraction methods and classification algorithms for the electroencephalographic (EEG) signals recorded in a motor task imagery paradigm. The sessions were performed by the same subject in eight consecutive years. Modeling the EEG signal as an autoregressive process (by means of Itakura distance and symmetric Itakura distance), amplitude modulation (using the amplitude modulation energy index) and phase synchronization (measuring phase locking value, phase lag index and weighted phase lag index) are the methods used for getting the appropriate information. The extracted features are classified using linear discriminant analysis, quadratic discriminant analysis, Mahalanobis distance, support vector machine and k nearest neighbor classifiers. The highest classifications rates are achieved when Itakura distance with Mahalanobis distance based classifier are app...
Information Technology And Control, 2019
The motor imagery (MI) based brain-computer interface systems (BCIs) can help with new communication ways. A typical electroencephalography (EEG)-based BCI system consists of several components including signal acquisition, signal pre-processing, feature extraction and feature classification. This paper focuses on the feature extraction step and proposes to use a combination of different feature extraction and feature reduction methods. The research presented in the paper explores the methods of band power, time domain parameters, fast Fourier transform and channel variance for feature extraction. These methods are investigated by combining them in pairs. The application of two feature extraction methods increases the number of selected features that can be redundant or irrelevant. The utilization of too many features can lead to wrong classification results. Therefore, the methods of feature reduction have to be applied. The following feature reduction methods are investigated: pri...
A Feature Extraction Scheme to Classify Motor Imagery MovementsBased on Bi-spectrum Analysis of EEG
2016
In this paper, an effective but simple feature extraction procedure for motor imagery movement classification has been proposed. Higher-order statistical features are extracted using bi-spectrum analysisof the non-linear EEG signal. Firstly, EEG signals are filtered through a band-pass filter for decomposing EEG signal into several bands and then these signals are bi-spected. Higher-order statistical features i.e. skewness, kurtosis, V2 order, V3 orders, Variance etc. are investigated. From the one-way ANOVA analysis, these features are shown to be promising to distinguish motor imagery hand movement of EEG signals. The whole experiment accomplished by using the publically available benchmark BCI-competition 2003 Graz motor imagery dataset. Different types of classifiers have been tested to classify EEG signal, among themK-Nearest Neighbors (KNN) classifier provides a good accuracy of 84.29%. Finally, proposed method is compared with some of the existing methods and superior perform...
Nonlinear and nonstationary framework for feature extraction and classification of motor imagery
IEEE ... International Conference on Rehabilitation Robotics : [proceedings], 2011
In this work we investigate a nonlinear approach for feature extraction of Electroencephalogram (EEG) signals in order to classify motor imagery for Brain Computer Interface (BCI). This approach is based on the Empirical Mode Decomposition (EMD) and band power (BP). The EMD method is a data-driven technique to analyze non-stationary and nonlinear signals. It generates a set of stationary time series called Intrinsic Mode Functions (IMF) to represent the original data. These IMFs are analyzed with the power spectral density (PSD) to study the active frequency range correspond to the motor imagery for each subject. Then, the band power is computed within a certain frequency range in the channels. Finally, the data is reconstructed with only the specific IMFs and then the band power is employed on the new database. The classification of motor imagery was performed by using two classifiers, Linear Discriminant Analysis (LDA) and Hidden Markov Models (HMMs). The results obtained show tha...
A Comparative Study of the EEG Characteristics for Motor Execution and Motor Imagery
Brain Computer Interface (BCI) has been improving the lifestyle of differently abled people by boosting up their performance levels. In this paper, Welch and Yule Walker-Power spectral density (PSD) have been used as a measure to differentiate various characteristics of EEG signal based on limb movements and their imagery. Characterization of partial limb movement has been performed in order to increase the flexibility of the BCI. The higher estimate for C3 for right hand movement shows contra-lateral activation of the Brain. Both Yule walker and Welch based methods were able to distinguish partial limb movements effectively in both C3 and C4 electrode data. Our approach presented in paper has shown good supporting results for execution and computation of robust features which can be utilized for signal classification.