Brain–Computer Interface: The HOL–SSA Decomposition and Two-Phase Classification on the HGD EEG Data
Abstract
:1. Introduction
2. Literature Survey
3. Methods and Materials
3.1. Singular Spectrum Analysis (SSA)
3.2. HOL–SSA
3.2.1. HOSVD
- SVD of m-mode unfolding of R as
- After the computation of matrices of m mode singular vectors , the core tensor can be computed as follows.
- Number of nonzero diagonal elements in as the m rank.
3.2.2. Truncated HOSVD
3.2.3. L-Moment
Algorithm 1: λ2 = (EX2:2 − EX1:2)/2λ3 = (EX3:3 − 2EX2:3+EX1:3)/3Proposed HOL–SSA And ORICA Methodology |
Step 1: Input the raw EEG signal. Step 2: Map the signal vector to a matrix. In the embedding stage, the time series s with length l is mapped into tensor , where s is segmented using a nonoverlapping window of size i and a[l/i] x i matrix M is obtained from s.
L = [l/i]
Because the application of SSA to real data does not exploit the inherent nonstationarity and therefore may fail in actual data decomposition, therefore, tensor-based SSA is a robust solution to this problem. Step 3: Decompose the signal using HOSVD. The truncated HOSVD of the converted tensor of order O and the dominant m ranks for is computed. for { compute compute matrices of dominant m-mode singular vectors } compute compute from Step 4: Determine the Linear moments of HOSVD. The nth population L-moment of a tensor with O order statistics in a decomposed sample from the distribution of core tensor is as follows. Step 5: Reconstruct the original signal to a multivariate data matrix. The matrices from step 4 are grouped into submatrices, as given below. Secondly, each matrix of the grouped decomposition is Hankelized, after which the Hankel matrix is transformed into a new series of length . The diagonal averaging applied to the resultant matrix produces a reconstructed series. Thus, the initial series set is decomposed into a sum of r reconstructed subseries, as shown below. Step 6: Apply ORICA on the multivariate data matrix, and for each iteration, the whitening matrix and the demixing matrix are computed. In order to reverse the mixing action, the inverse matrix of the reconstructed subseries is built. The independent components are produced by applying the ORICA rule after applying the Sherman–Morrison matrix inversion method. Step 7: Output the mapped sources of interest into original signal form. Time Complexity: |
4. Result and Analysis
4.1. Dataset Description
4.2. Performance Analysis
5. Conclusions
Author Contributions
Funding
Institutional Review Board Statement
Informed Consent Statement
Data Availability Statement
Acknowledgments
Conflicts of Interest
References
- Machado, S.; Araújo, F.; Paes, F.; Velasques, B.; Cunha, M.; Budde, H.; Basile, L.F.; Anghinah, R.; Arias-Carrión, O.; Cagy, M.; et al. EEG-based Brain-Computer Interfaces: An Overview of Basic Concepts and Clinical Applications in Neurorehabilitation. Rev. Neurosci. 2010, 21, 451–468. [Google Scholar] [CrossRef]
- Martinek, R.; Ladrova, M.; Sidikova, M.; Jaros, R.; Behbehani, K.; Kahankova, R.; Kawala-Sterniuk, A. Advanced Bioelectrical Signal Processing Methods: Past, Present and Future Approach—Part II: Brain Signals. Sensors 2021, 21, 5186. [Google Scholar] [CrossRef]
- Taha, L.Y.; Abdel-Raheem, E. Blind Source Separation: A Performance Review Approach. In Proceedings of the 5th International Conference on Signal Processing and Information Security (ICSPIS), Dubai, United Arab Emirates, 7–8 December 2022. [Google Scholar]
- Chaddad, A.; Wu, Y.; Kateb, R.; Bouridane, A. Electroencephalography Signal Processing: A Comprehensive Review and Analysis of Methods and Techniques. Sensors 2023, 23, 6434. [Google Scholar] [CrossRef]
- Wang, X.; Wang, X.; Zhou, B. EEG Artifact Removal Based on Independent Component Analysis and Outlier Detection. In Proceedings of the 2022 14th International Conference on Wireless Communications and Signal Processing (WCSP), Nanjing, China, 14–17 October 2022; pp. 209–214. [Google Scholar] [CrossRef]
- Taelman, J.; Mijovic, B.; Van Huffel, S.; Devuyst, S.; Dutoit, T. ECG artifact removal from surface EMG signals by combining empirical mode decomposition and independent component analysis. In Proceedings of the 4th International Conference on Bio-inspired Systems and Signal Processing, Rome, Italy, 26–29 January 2011; pp. 421–424. [Google Scholar]
- Blokland, Y.; Bruhn, J.; Farquhar, J.; Spyrou, L. Singular spectrum analysis as a preprocessing filtering step for fNIRS brain computer interfaces. In Proceedings of the 22nd European Signal Processing Conference, Lisbon, Portugal, 1–5 September 2014; pp. 46–50. [Google Scholar]
- Maddirala, A.K.; Shaik, R.A. Separation of Sources from Single-Channel EEG Signals Using Independent Component Analysis. IEEE Trans. Instrum. Meas. 2017, 67, 382–393. [Google Scholar] [CrossRef]
- Gopalan, G.; Wikle, C.K. A Higher-Order Singular Value Decomposition Tensor Emulator for Spatiotemporal Simulators. J. Agric. Biol. Environ. Stat. 2022, 27, 22–45. [Google Scholar] [CrossRef]
- Hassani, H.; Yeganegi, M.R.; Silva, E.S. A New Signal Processing Approach for Discrimination of EEG Recordings. Stats 2018, 1, 155–168. [Google Scholar] [CrossRef]
- Yedukondalu, J.; Sharma, L.D. Circulant Singular Spectrum Analysis and Discrete Wavelet Transform for Automated Removal of EOG Artifacts from EEG Signals. Sensors 2023, 23, 1235. [Google Scholar] [CrossRef] [PubMed]
- Liu, Q.; Liu, A.; Zhang, X.; Chen, X.; Qian, R.; Chen, X. Removal of EMG Artifacts from Multichannel EEG Signals Using Combined Singular Spectrum Analysis and Canonical Correlation Analysis. J. Healthc. Eng. 2019, 2019, 4159676. [Google Scholar] [CrossRef]
- Chao-Lin, T.; Yi-Yang, Z.; Wei, W.; Yuan-Yuan, L.; Gang, W.; Jin, X. A Novel Method Based on Combination of Independent Component Analysis and Ensemble Empirical Mode Decomposition for Removing Electrooculogram Artifacts from Multichannel Electroencephalogram Signals. Front. Neurosci. 2021, 15, 729403. [Google Scholar]
- Sandoval, C.; Pirogova, E.; Lech, M. Two-Stage Deep Learning Approach to the Classification of Fine-Art Paintings. IEEE Access 2019, 7, 41770–41781. [Google Scholar] [CrossRef]
- Naveed, M.; Arif, F.; Usman, S.M.; Anwar, A.; Hadjouni, M.; Elmannai, H.; Hussain, S.; Ullah, S.S.; Umar, F. A Deep Learning-Based Framework for Feature Extraction and Classification of Intrusion Detection in Networks. Wirel. Commun. Mob. Comput. 2022, 2022, 2215852. [Google Scholar] [CrossRef]
- Demidova, L.A. Two-Stage Hybrid Data Classifiers Based on SVM and kNN Algorithms. Symmetry 2021, 13, 615. [Google Scholar] [CrossRef]
- Yang, Y.; Hu, Y.; Zhang, X.; Wang, S. Two-Stage Selective Ensemble of CNN via Deep Tree Training for Medical Image Classification. IEEE Trans. Cybern. 2022, 52, 9194–9207. [Google Scholar] [CrossRef]
- Dos Santos, E.M.; Cassani, R.; Falk, T.H.; Fraga, F.J. Improved motor imagery brain-computer interface performance via adaptive modulation filtering and two-stage classification. Biomed. Signal Process. Control. 2020, 57, 101812. [Google Scholar] [CrossRef]
- Malekzadeh, A.; Zare, A.; Yaghoobi, M.; Kobravi, H.-R.; Alizadehsani, R. Epileptic Seizures Detection in EEG Signals Using Fusion Handcrafted and Deep Learning Features. Sensors 2021, 21, 7710. [Google Scholar] [CrossRef] [PubMed]
- Malekzadeh, A.; Zare, A.; Yaghoobi, M.; Alizadehsani, R. Automatic Diagnosis of Epileptic Seizures in EEG Signals Using Fractal Dimension Features and Convolutional Autoencoder Method. Big Data Cogn. Comput. 2021, 5, 78. [Google Scholar] [CrossRef]
- Shen, Y.; Peng, F.; Li, B. Improved singular spectrum analysis for time series with missing data. Nonlinear Process. Geophys. 2014, 22, 371–376. [Google Scholar] [CrossRef]
- Xu, S.; Hu, H.; Ji, L.; Wang, P. Embedding Dimension Selection for Adaptive Singular Spectrum Analysis of EEG Signal. Sensors 2018, 18, 697. [Google Scholar] [CrossRef]
- Das, S.; Kyal, C. Efficient multichannel EEG compression by optimal tensor truncation. Biomed. Signal Process. Control. 2021, 68, 102749. [Google Scholar] [CrossRef]
- Zhang, L.-W.; Lu, Z.-H.; Zhao, Y.-G. Dynamic reliability assessment of nonlinear structures using extreme value distribution based on L-moments. Mech. Syst. Signal Process. 2021, 159, 107832. [Google Scholar] [CrossRef]
- Antony, M.J.; Sankaralingam, B.P.; Mahendran, R.K.; Gardezi, A.A.; Shafiq, M.; Choi, J.-G.; Hamam, H. Classification of EEG Using Adaptive SVM Classifier with CSP and Online Recursive Independent Component Analysis. Sensors 2022, 22, 7596. [Google Scholar] [CrossRef] [PubMed]
- Judith, A.M.; Priya, S.B.; Mahendran, R.K.; Gadekallu, T.R.; Ambati, L.S. Two-phase classification: ANN and A-SVM classifiers on motor imagery BCI. Asian J. Control. 2022, 1–12, 1561–8625. [Google Scholar]
- Schirrmeister, R.T.; Springenberg, J.T.; Fiederer, L.D.J.; Glasstetter, M.; Eggensperger, K.; Tangermann, M.; Hutter, F.; Burgard, W.; Ball, T. Deep learning with convolutional neural networks for EEG decoding and visualization. Hum. Brain Mapp. 2017, 38, 5391–5420. [Google Scholar] [CrossRef]
- Lawhern, V.J.; Solon, A.J.; Waytowich, N.R.; Gordon, S.M.; Hung, C.P.; Lance, B.J. EEGNet: A compact convolutional neural network for EEG-based brain–computer interfaces. J. Neural Eng. 2018, 15, 056013. [Google Scholar] [CrossRef] [PubMed]
- Li, Y.; Zhang, X.-R.; Zhang, B.; Lei, M.-Y.; Cui, W.-G.; Guo, Y.-Z. A Channel-Projection Mixed-Scale Convolutional Neural Network for Motor Imagery EEG Decoding. IEEE Trans. Neural Syst. Rehabilitation Eng. 2019, 27, 1170–1180. [Google Scholar] [CrossRef]
- Altuwaijri, G.A.; Muhammad, G. Electroencephalogram-Based Motor Imagery Signals Classification Using a Multi-Branch Convolutional Neural Network Model with Attention Blocks. Bioengineering 2022, 9, 323. [Google Scholar] [CrossRef]
- Haq A u Li, J.P.; Khan, S.; Alshara, M.A.; Alotaibi, R.M.; Mawuli, C.B. DACBT: Deep learning approach for classifica-tion of brain tumors using MRI data in IoT healthcare environment. Sci. Rep. 2022, 12, 15331. [Google Scholar] [CrossRef]
- Haq, A.U.; Li, J.P.; Agbley, B.L.Y.; Khan, A.; Khan, I.; Uddin, M.I.; Khan, S. IIMFCBM: Intelligent Integrated Model for Feature Extraction and Classification of Brain Tumors Using MRI Clinical Imaging Data in IoT-Healthcare. IEEE J. Biomed. Health Inform. 2022, 26, 5004–5012. [Google Scholar] [CrossRef]
- Salehi, A.W.; Khan, S.; Gupta, G.; Alabduallah, B.I.; Almjally, A.; Alsolai, H.; Siddiqui, T.; Mellit, A. A Study of CNN and Transfer Learning in Medical Imaging: Advantages, Challenges, Future Scope. Sustainability 2023, 15, 5930. [Google Scholar] [CrossRef]
Components/Signals | EEG (%) | Muscle (%) | Eye (%) | Heart (%) | Line Noise (%) | Channel Noise (%) | Other (%) |
---|---|---|---|---|---|---|---|
IC 1 | 97.1 | 4.6 | 0.0 | 0.1 | 0.3 | 0.4 | 3.0 |
IC 2 | 10.0 | 85.2 | 0.6 | 0.0 | 1.4 | 0.1 | 2.6 |
IC 3 | 3.3 | 93.3 | 1.0 | 0.0 | 0.8 | 0.0 | 1.6 |
IC 4 | 1.6 | 95.9 | 0.6 | 0.0 | 0.6 | 0.0 | 1.2 |
IC 5 | 2.1 | 95.2 | 0.7 | 0.0 | 0.6 | 0.0 | 1.3 |
IC 6 | 0.9 | 84.7 | 9.0 | 0.0 | 0.3 | 0.0 | 5.0 |
IC 7 | 0.6 | 66.7 | 22.4 | 0.0 | 1.2 | 0.1 | 9.0 |
IC 8 | 0.2 | 73.0 | 5.3 | 0.0 | 0.4 | 0.3 | 20.9 |
IC 9 | 0.9 | 93.5 | 0.4 | 0.0 | 0.8 | 0.1 | 4.3 |
IC 10 | 6.8 | 50.6 | 1.3 | 0.4 | 8.2 | 0.3 | 32.2 |
IC 11 | 7.6 | 83.7 | 1.2 | 0.3 | 1.2 | 0.1 | 5.9 |
IC 12 | 12.7 | 77.4 | 0.2 | 3.3 | 1.2 | 0.1 | 5.2 |
IC 13 | 14.0 | 40.1 | 0.9 | 0.2 | 5.3 | 3.2 | 36.4 |
IC 14 | 12.3 | 63.0 | 0.7 | 1.1 | 3.5 | 0.1 | 19.2 |
IC 15 | 1.0 | 37.9 | 13.2 | 0.0 | 0.3 | 0.4 | 47.2 |
IC 16 | 1.3 | 90.1 | 1.9 | 0.0 | 0.8 | 0.3 | 5.6 |
IC 17 | 2.4 | 60.2 | 3.1 | 0.1 | 0.2 | 0.6 | 33.5 |
IC 18 | 5.7 | 47.7 | 2.8 | 0.5 | 1.5 | 0.6 | 44.2 |
IC 19 | 1.9 | 67.7 | 1.1 | 0.0 | 0.4 | 0.8 | 28.1 |
IC 20 | 0.7 | 82.2 | 0.9 | 0.2 | 0.3 | 0.8 | 15.1 |
IC 21 | 0.3 | 91.1 | 1.9 | 0.0 | 0.0 | 0.5 | 6.1 |
Participants | S1 | S2 | S3 | S4 | S5 | S6 | S7 | S8 | S9 | S10 | S11 | S12 | S13 | S14 | Avg. | |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Accuracy (%) | 94.2 | 96.4 | 94.4 | 95.6 | 93.9 | 96.2 | 96.7 | 94.2 | 94.9 | 95.2 | 96.9 | 96.5 | 94.6 | 93.7 | 95.2 | |
K value | 0.92 | 0.95 | 0.93 | 0.94 | 0.92 | 0.95 | 0.95 | 0.92 | 0.93 | 0.94 | 0.96 | 0.95 | 0.93 | 0.92 | 0.94 | |
Precision | LH | 0.93 | 0.97 | 0.93 | 0.96 | 0.93 | 0.97 | 0.96 | 0.93 | 0.95 | 0.96 | 0.97 | 0.96 | 0.95 | 0.93 | 0.95 |
RH | 0.94 | 0.97 | 0.94 | 0.97 | 0.93 | 0.96 | 0.97 | 0.94 | 0.94 | 0.96 | 0.97 | 0.97 | 0.95 | 0.93 | 0.95 | |
FT | 0.95 | 0.96 | 0.95 | 0.96 | 0.95 | 0.96 | 0.98 | 0.96 | 0.95 | 0.96 | 0.98 | 0.97 | 0.96 | 0.94 | 0.96 | |
RT | 0.95 | 0.96 | 0.96 | 0.94 | 0.95 | 0.96 | 0.96 | 0.94 | 0.96 | 0.93 | 0.96 | 0.96 | 0.93 | 0.95 | 0.95 | |
Avg | 0.94 | 0.97 | 0.95 | 0.96 | 0.94 | 0.96 | 0.97 | 0.94 | 0.95 | 0.95 | 0.97 | 0.97 | 0.95 | 0.94 | 0.95 | |
Recall | LH | 0.94 | 0.95 | 0.94 | 0.95 | 0.93 | 0.96 | 0.96 | 0.95 | 0.94 | 0.93 | 0.97 | 0.96 | 0.93 | 0.93 | 0.95 |
RH | 0.94 | 0.96 | 0.94 | 0.96 | 0.94 | 0.95 | 0.97 | 0.94 | 0.95 | 0.96 | 0.97 | 0.97 | 0.94 | 0.93 | 0.95 | |
FT | 0.95 | 0.97 | 0.95 | 0.97 | 0.94 | 0.97 | 0.98 | 0.94 | 0.95 | 0.96 | 0.98 | 0.97 | 0.97 | 0.94 | 0.96 | |
RT | 0.94 | 0.97 | 0.95 | 0.95 | 0.95 | 0.97 | 0.96 | 0.94 | 0.96 | 0.96 | 0.96 | 0.96 | 0.95 | 0.95 | 0.96 | |
Avg | 0.94 | 0.96 | 0.95 | 0.96 | 0.94 | 0.96 | 0.97 | 0.94 | 0.95 | 0.95 | 0.97 | 0.97 | 0.95 | 0.94 | 0.95 | |
F1-score | LH | 0.94 | 0.96 | 0.93 | 0.95 | 0.93 | 0.96 | 0.96 | 0.94 | 0.94 | 0.94 | 0.97 | 0.96 | 0.94 | 0.93 | 0.95 |
RH | 0.94 | 0.96 | 0.94 | 0.96 | 0.93 | 0.95 | 0.97 | 0.94 | 0.94 | 0.96 | 0.97 | 0.97 | 0.94 | 0.93 | 0.95 | |
FT | 0.95 | 0.97 | 0.95 | 0.96 | 0.94 | 0.96 | 0.98 | 0.95 | 0.95 | 0.96 | 0.98 | 0.97 | 0.96 | 0.94 | 0.96 | |
RT | 0.94 | 0.96 | 0.95 | 0.94 | 0.95 | 0.96 | 0.96 | 0.94 | 0.96 | 0.94 | 0.96 | 0.96 | 0.94 | 0.95 | 0.95 | |
Avg | 0.94 | 0.96 | 0.94 | 0.95 | 0.94 | 0.96 | 0.97 | 0.94 | 0.95 | 0.95 | 0.97 | 0.97 | 0.95 | 0.94 | 0.95 | |
Misclassification Rate | 0.058 | 0.036 | 0.056 | 0.044 | 0.06 | 0.038 | 0.033 | 0.058 | 0.05 | 0.048 | 0.031 | 0.035 | 0.054 | 0.063 | 0.0476 |
Methods/Subjects | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | Avg. |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
DeepConvNet | 81.88 | 91.88 | 93.13 | 92.50 | 90.63 | 93.13 | 84.28 | 90.80 | 96.88 | 85.00 | 88.13 | 91.25 | 89.94 | 83.75 | 89.51 |
EEGNet | 94.37 | 92.50 | 100 | 96.25 | 96.87 | 98.12 | 93.07 | 96.87 | 98.12 | 91.25 | 80.00 | 96.25 | 95.60 | 79.37 | 93.47 |
CP-MixedNet | 88.75 | 90.00 | 95.63 | 91.25 | 95.00 | 91.25 | 88.05 | 93.13 | 95.00 | 88.75 | 75.63 | 93.75 | 89.31 | 78.13 | 93.70 |
TS-SEFFNet | 90.69 | 93.53 | 98.53 | 96.88 | 92.90 | 93.53 | 92.40 | 91.78 | 96.88 | 89.88 | 92.78 | 95.40 | 93.03 | 87.34 | 93.25 |
MBEEGNet | 95.02 | 95.02 | 100 | 99.40 | 98.17 | 98.80 | 93.13 | 95.52 | 98.18 | 92.14 | 89.43 | 96.02 | 94.45 | 88.88 | 95.30 |
MBShallowCovNet | 98.25 | 96.23 | 98.80 | 98.18 | 97.65 | 96.90 | 93.80 | 97.00 | 97.52 | 92.50 | 80.78 | 96.25 | 95.62 | 92.04 | 95.11 |
ANN + A-SVM | 94.2 | 96.4 | 94.4 | 95.6 | 93.9 | 96.2 | 96.7 | 94.2 | 94.9 | 95.2 | 96.9 | 96.5 | 94.6 | 93.7 | 95.24 |
Dataset | Methods | Accuracy (%) | F1-Score | Reference | |
---|---|---|---|---|---|
Feature Extraction | Classification | ||||
HGD | ShallowConvNet | CNN | 88.69 | 0.887 | Schirrmeister, et.al [27] (2017) |
DeepConvNet | CNN | 89.51 | 0.893 | Schirrmeister, et.al [27] (2017) | |
EEGNet | CNN | 93.47 | 0.935 | Lawhern, et al. [28] (2018) | |
CP-MixedNet | CNN | 93.70 | 0.937 | Li, et.al [29] (2019) | |
TS-SEFFNet | CNN | 93.25 | 0.901 | Li, et.al [29] (2019) | |
MBEEGNet | CNN | 95.30 | 0.954 | Altuwaijri and Muhammad [30] (2022) | |
MBShallowCovNet | CNN | 95.11 | 0.951 | Altuwaijri and Muhammad [30,31,32,33] (2022) | |
HOL–SSA–ORICA + CSP | ANN + A-SVM | 95.24 | 0.95 | - |
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Antony, M.J.; Sankaralingam, B.P.; Khan, S.; Almjally, A.; Almujally, N.A.; Mahendran, R.K. Brain–Computer Interface: The HOL–SSA Decomposition and Two-Phase Classification on the HGD EEG Data. Diagnostics 2023, 13, 2852. https://doi.org/10.3390/diagnostics13172852
Antony MJ, Sankaralingam BP, Khan S, Almjally A, Almujally NA, Mahendran RK. Brain–Computer Interface: The HOL–SSA Decomposition and Two-Phase Classification on the HGD EEG Data. Diagnostics. 2023; 13(17):2852. https://doi.org/10.3390/diagnostics13172852
Chicago/Turabian StyleAntony, Mary Judith, Baghavathi Priya Sankaralingam, Shakir Khan, Abrar Almjally, Nouf Abdullah Almujally, and Rakesh Kumar Mahendran. 2023. "Brain–Computer Interface: The HOL–SSA Decomposition and Two-Phase Classification on the HGD EEG Data" Diagnostics 13, no. 17: 2852. https://doi.org/10.3390/diagnostics13172852
APA StyleAntony, M. J., Sankaralingam, B. P., Khan, S., Almjally, A., Almujally, N. A., & Mahendran, R. K. (2023). Brain–Computer Interface: The HOL–SSA Decomposition and Two-Phase Classification on the HGD EEG Data. Diagnostics, 13(17), 2852. https://doi.org/10.3390/diagnostics13172852