Table of contentsIntroductionData acquisitionDimensionality reduction techniqueFeature classification using SVVMConclusionThis paper proposes a modified algorithm for functional classification of hand imaging movements left and right obtained from the EEG signal. Electroencephalogram (EEG) is the signal acquired from the human brain to monitor and identify human actions in the face of different stimuli. The data were obtained from the BCI III (b) 2003 competition, acquired by the Graz University of Technology. The recorded EEG was sampled at 125 Hz and filtered between 0.5 and 30 Hz. Say no to plagiarism. Get a tailor-made essay on “Why Violent Video Games Should Not Be Banned”? Get the original essay Features were extracted using discrete wavelet transform (DWT). To obtain accurate detailed information, the EEG signal was processed with dimensionality reduction techniques used as (i) singular value decomposition and (ii) LDA. Support vector machines (SVM) were used for optimal classification of each motor movement. The result for the binary SVM class was at the 100% accuracy level. The results established the accuracy of singular value decomposition as the best tool for identifying imaging movements. Introduction Feature extraction and feature classification have always been the difficult task in EEG signal. The EEG signal provides detailed information about the electrical activity in the brain. It offers an alternative form of communication to people with disabilities. Our work focused on reducing the complexity of the information and, on the other hand, retaining vital information from the placement of the C3 and C4 electrodes. C3 and C4 are an integral part of the transmission of sensorimetric information from the brain. The EEG is obtained from 10 to 20 electrode placements conforming to international standards [11] on the surface of the skull. Positions C3 and C4 are the regions that provide theta rhythms. In the proposed work, the motor imagery movements of the left and right hands were classified. The extensive research on feature extraction and feature classification has been presented with great success. But managing complexity remains the major problem in EEG signal classification. Xiao-Dong ZHANG, et.al [2] had presented the prosthetic hand control algorithm. The EEG signal was analyzed based on several complicated hand movements. The author concluded that the classification obtained from support vector machines was much better than that obtained with ANN. Andrews S. et al [21] had presented singular value decomposition (SVD) to reduce noise and data dimension. The experimental results yielded a very low false acceptance rate (FAR) and false rejection rate (FRR) and an almost negligible equal error rate (EER) of 2.91%. Sachin Garg et al [22] demonstrated the use of wavelet transform for feature extraction. of the EEG signal. The author claimed that after extracting the coefficients, it was remarkably easier to calculate the statistical parameters of the EEG signal. Another author, Ashwini Nakate et al [24] had also advocated the use of discrete wavelet transform technique to decompose the EEG signal. Priyanka Khatwani et al [25] present the DWT technique to denoise EEG signal data. Rajesh Singla et.al [26] presented the motor imagery movement of the wrist, clockwise/counterclockwise rotation of the wrist of the elbow and theankle backward/forward. It has been argued that the DWT technique is best suited for extracting features from the EEG signal. Abdulhamit Subasi et.al [27] presented the comparison between different techniques used to manipulate EEG signal data. Principal component analysis (PCA), independent component analysis (ICA), and linear discriminant analysis (LDA) were used to reduce the dimensionality of the signal. Siuly et. [28] proposed a statistical algorithm to correctly classify the function of the EEG signal. Thanh and. Al [29] had analyzed the EEG signals using the type 2 fuzzy logic method. The results had shown the low computational cost with good accuracy. ABM Hossain et al [30] had proposed the probabilistic neural network algorithm for optimal functional classification of EEG signal. The author claimed an accuracy rate of approximately 99.7%. The work of MA Hassan et al [31] focused on modifying the back-propagation neural network for the EEG signal. The classification rate was between 97 and 100%. It also concluded that time domain features extracted from EEG were more reliable for feature classification. The paper is organized as follows (Figure 1.1): Section II is data acquisition. It processes detailed database information of left and right hand imaging movements. Section III focuses on feature extraction using DWT. Section IV discusses dimensionality reduction techniques. Additionally, Section V deals with the identification of motor movements. Finally, Section VI validates our proposed algorithm. While section VII concludes the work. Data acquisition The database was obtained from Graz Univ. of technology (BCI III(b) competition, 2003). The imaging signals and movements of the left and right hands were preprocessed to remove artifacts from various noises (biosignals/external). Three electrodes (C3, Cz and C4) were placed to record the EEG data with a sampling frequency of 125 Hz. Band-pass filters were used with a frequency range between 0.5 Hz and 30 Hz, and a 50 Hz notch filter was also used to remove artifacts. The dataset was recorded from a normal subject (female, 25 years old). The subject did not receive any information about the recording. The comfortable chair with armrests was provided. The task consisted of acquiring imaginary movements of the left/right hand. The experiment includes 7 trials of 40 trials each. Each trial lasts 9 seconds. After an initial rest of 2 seconds, the recording of the respective motor movements was started. Trials were then selected for random training and testing to classify the imagery movements. In our proposed work, C3 and C4 (electrode placement) were considered for further analysis.[16] Figure 1.3 is the flowchart of feature extraction using DWT. The three motor movements were dispensed to refine the coefficients using Symlet at decomposition level “3” [38], so no useful information should be diminished. Using Symlet, the extracted features were almost symmetric and had the least asymmetry. The associated scaling filters are near-linear phase filters. Dimensionality Reduction Technique For further analysis, dimensionality reduction, singular value decomposition (SVD) and linear discriminant analysis (LDA) were implemented to increase the computational efficiency of the proposed algorithm . It was found useful to remove all unrelated and redundant features from.