Extracting suitable features for detecting voluntary movements from EEG signals
Subject Areas :
parastoo ghafourpoor
1
,
Reza Ghafouri
2
*
,
omid mahdiyar
3
1 -
2 - Department of Electrical Engineering, Kazeroun Branch, Islamic Azad University, Kazeroun, Iran
3 - Department of electrical and computer engineering, kazerun branch, Islamic azad university, kazerun, iran
Keywords: EEG signals, voluntary movements, feature extraction, brain signal recognition, Walsh transform, signal entropy.,
Abstract :
This research focuses on extracting suitable features for detecting voluntary movements from EEG signals. To achieve this, a method with three main stages was proposed: noise removal, feature extraction, and classification. In the first step, brain signals were cleaned of recording device noise using four methods. These included two classic techniques (Independent Component Analysis - ICA and Wavelet Transform) and two newer approaches (Walsh Transform and a combination of Walsh Transform with ICA). The performance of these methods was evaluated using three criteria: Signal-to-Noise Ratio (SNR), Mean Squared Error (MSE), and Percentage Root Mean Square Difference (PRD). The results showed that the combined Walsh and ICA method, as well as the Walsh Transform alone, performed better than other methods by yielding the highest SNR values and the lowest MSE and PRD values. In the feature extraction stage, vectors consisting of 22 components were calculated for various channels and the entire signal. These components included entropy and power from Walsh Transform. For comparison, similar features were also extracted using the Wavelet and Fourier Transforms. Finally, data classification was performed using two methods: Support Vector Machine (SVM) and K-Nearest Neighbors (KNN). The results indicate that Walsh Transform-based features provided the best detection rates. The classification accuracy was 42.5% with SVM and 39.0% with KNN. Furthermore, evaluation with BCI Competition IV data showed that the proposed method, except in one instance, outperformed other approaches and demonstrated excellent performance in terms of training and testing time. Its average processing time was 52 seconds, which is considerably less than the comparative methods (403 and 640 seconds).
[1] R. G. Babu, P. Karthika, K. Elangovan, “Performance analysis for image security using SVM and ANN classification techniques”, in Proc. 3rd Int. Conf. Electron., Commun. Aerosp. Technol. (ICECA), pp. 460–466, 2019.
[2] A. Swetapadma, A. Yadav, A. Y. Abdelaziz, “Intelligent schemes for fault classification in mutually coupled series-compensated parallel transmission lines”, Neural Comput. Appl., vol. 32, no. 11, pp. 6939–6956, 2020.
[3] BCI Competition IV - Dataset Results. Available at: http://www.bbci.de/competition/iv/, Accessed: May 5, 2025.
[4] A. Hyvarinen, J. Karhunen, E. Oja, Independent Component Analysis - Theory and Applications. New York, NY, USA: Springer, 2010.
[5] G. Baghdadi, F. Hadaeghi, C. Kamarajan, “Multimodal approaches to investigating neural dynamics in cognition and related clinical conditions: Integrating EEG, MEG, and fMRI data”, Front. Syst. Neurosci., vol. 19, pp. 149–158, 2025.
[6] S. A. Hosseini, M. Houshmand, “Analysis of the EEG Signal Using Higher-Order Spectra (HOS) in the Neuro-marketing Application”, New Marketing Research Journal, vol. 12, no. 1, pp. 25–42, 2022.
[7] S. A. Othman, K. M. Omar, “An enhanced shrinkage function for denoising economic time series data using wavelet analysis”, Sci. J. Univ. Zakho, vol. 12, no. 1, pp. 138–143, 2024.
[8] N. Boukhennoufa, Y. Laamari, R. Benzid, “Signal denoising using a low computational translation-invariant-like strategy involving multiple wavelet bases: Application to synthetic and ECG signals”, Metrol. Meas. Syst., pp. 259–278, 2024.
[9] S. M. Najeeb, H. T. Al Rikabi, S. M. Ali, “Finding the discriminative frequencies of motor electroencephalography signal using genetic algorithm”, TELKOMNIKA (Telecommun. Comput. Electron. Control), vol. 19, no. 1, pp. 285–292, 2021.
[10] T. F. Zaidi, O. Farooq, “EEG sub-bands based sleep stages classification using Fourier Synchrosqueezed transform features”, Expert Syst. Appl., vol. 212, p. 118752, 2023.