Brain Computer Interface / BCI / Neural Control Int. / NCI / Mind Machine Int. / MMI / Direct Neural Int. / DNI / Brain Machine Int. / BMI
maryam farhadnia; Sepideh Hajipour; mohammad mikaili
Volume 17, Issue 1 , May 2023, , Pages 1-10
Abstract
Today, usage of brain-computer interface systems based on steady-state visual evoked potentials (SSVEPs) has been increased due to some advantages such as acceptable accuracy and minimal need for user training. Steady-state visual potentials are one of the most important patterns used in BCI systems, ...
Read More
Today, usage of brain-computer interface systems based on steady-state visual evoked potentials (SSVEPs) has been increased due to some advantages such as acceptable accuracy and minimal need for user training. Steady-state visual potentials are one of the most important patterns used in BCI systems, which are generated in the occipital region of the brain by visual stimulation between 6 and 60 Hz. One of the effective methods for extracting the SSVEP frequency in BCI systems is called the Multiway Correlation Coefficient Analysis (MCCA) method, which is a tensorized version of the classical Correlation Coefficient Analysis (CCA) method and is based on multidimensional data.In this paper, inspired by the MCCA method, two new algorithms (PARAFAC-CCA and C-PARAFAC-CCA) have been proposed using the combination of CCA and PARAFAC decomposition. The purpose of the proposed algorithms is to improve the initial reference signal and achieve higher accuracy in SSVEP frequency detection in BCI systems. In the PARAFAC-CCA algorithm, after performing the PARAFAC decomposition on the multidimensional training data and obtaining the time component, the CCA method is implemented between the obtained time component and the sine-cosine reference signal, and the optimal reference signal is made from its output. Finally, the MLR algorithm is used between the EEG test data and the optimal reference signal in order to achieve the target frequency. The general steps of the C-PARAFAC-CCA algorithm are also similar to PARAFAC-CCA, with the difference that in the calculation of the time component, constrained PARAFAC is used in such a way that in each step of the ALS algorithm, CCA is applied once and the time component is improved. The efficiency of the proposed algorithms was investigated on the real data set and it was shown that compared to the MCCA method, the proposed algorithms have reached a higher average accuracy.
Biomedical Signal Processing / Medical Signal Processing / Biosignal Processing
Mohammad Davood Khalili; Vahid Abootalebi; Hamid Saeedi-Sourck
Volume 16, Issue 1 , May 2022, , Pages 75-94
Abstract
The human brain is one of the most complex and heterogeneous networks, and brain signals contain a lot of information, so researchers in this field are always looking for proper solutions to select meaningful features and reduce the dimension of this information appropriately to lead to better classification. ...
Read More
The human brain is one of the most complex and heterogeneous networks, and brain signals contain a lot of information, so researchers in this field are always looking for proper solutions to select meaningful features and reduce the dimension of this information appropriately to lead to better classification. Two of the new tools for brain signal processing are Graph Signal Processing (GSP) and Meta-heuristic and Evolutionary methods. In this paper, a geometric structure and a mixed structure are considered for the brain graph and the weights of the edges in the mixed structure are calculated by a combination of two measures: geometric distance and correlation. To reduce the graph dimension, the weighted degree metric and a combination of the Kron reduction method and Graph Fourier Transform (KG) are used to properly preserve the information of all vertices of the graph into the selected vertices. Feature extraction is performed by Ledoit-Wolf shrinkage estimation and Tangent Space Mapping (TSM) method. For dimension reduction of extracted features, Principal Component Analysis (PCA) method and feature selection based on Differential Evolution (DE) are used. The selected features are given to several well-known machine learning classifiers. To evaluate the performance of the proposed method, dataset IVa from BCI Competition III has been used. The results show that the average classification accuracy of the proposed KG-PCA method with SVM-RBF and DT classifiers, in the structural graph and the functional-structural graph, is higher than the TSM-GFT method expressed in previous studies, and the DT classifier has achieved an average accuracy of 91.15±1.17. Also, according to the obtained results, the performance of the proposed KG-DE method has been better compared to KG-PCA and in the best case, the average accuracy of the SVM-RBF classifier is equal to 95.50±1.27.