EEG Signal Classification under RF Exposure Using Synthetic Data–Augmented Feedforward Neural Network

Muhammad Rasyid Rosli; Roshakimah Mohd Isa; Nur Irsalina Huda Nazri

doi:10.31436/iiumej.v27i3.4079

Authors

Muhammad Rasyid Rosli Universiti Teknologi MARA
Roshakimah Mohd Isa Universiti Teknologi MARA https://orcid.org/0009-0007-8797-1000
Nur Irsalina Huda Nazri Universiti Teknologi MARA

DOI:

https://doi.org/10.31436/iiumej.v27i3.4079

Keywords:

Artificial Neural Networks, eeg signals, natural synthetic, Classification

Abstract

Electroencephalogram (EEG) signal classification plays a critical role in understanding brain activity under varying radiofrequency (RF) exposure conditions. The demand for improving EEG classification modeling is increasing due to the growing number of gadgets emitting RF exposure. Current Electroencephalogram (EEG) classification methods struggle to process signals accurately because of insufficient data and limited variety in the datasets used to train models. This study aims to improve Electroencephalogram (EEG) signal classification using synthetic data and a Feedforward Neural Network (FNN). The goal is to analyze brain signals under different RF exposure conditions and increase classification accuracy using asymmetry ratio analysis and synthetic data for 3 exposure groups. Conditional Generative Adversarial Networks (CTGAN) are employed to generate 88 synthetic samples from an original dataset of 97 EEG recordings, stratified by exposure type and gender. The two sub-bands from the original data were used to generate synthetic data, divided into 3 groups: Left Exposure (LE), Right Exposure (RE), and Sham Exposure (SE). Each group was further split into male and female categories. Statistically significant variations in alpha- and beta-band activity are observed during RF exposure, validating the discriminative capability of the extracted features. Including synthetic data significantly improves classification performance, particularly during male exposure sessions, where the Feedforward Neural Network (FNN) achieves 98% training accuracy. The findings demonstrate that synthetic data augmentation is an effective strategy for improving Electroencephalogram (EEG) classification reliability and offers strong potential for radiofrequency (RF) exposure analysis and related biomedical applications.

ABSTRAK: Kajian ini bertujuan menambah baik klasifikasi isyarat Elektroensefalogram (EEG) bagi aktiviti otak di bawah pendedahan frekuensi radio (RF) menggunakan data sintetik dan Rangkaian Nueral Suap-Depan (FNN). Isu utama adalah kekurangan data dalam set data EEG sedia ada. Kami menggunakan analisis nisbah asimetri bagi mengekstrak ciri dan Analisis Varians (ANOVA) dan membandingkan gelombang alfa dan beta antara lelaki dan wanita dalam tiga kumpulan pendedahan: Kiri (LE), Kanan (RE), dan Palsu (SE). Rangkaian Berseteru Generatif Bersyarat (CTGAN) digunakan bagi menjana 88 data sintetik, meningkatkan set data asal 97 sampel. Penggunaan data sintetik meningkatkan ketepatan klasifikasi model, terutama untuk sesi lelaki, mencapai 98% ketepatan latihan dan 69% ketepatan ujian. Perubahan signifikan dalam gelombang alfa dan beta diperhatikan semasa pendedahan RF. Penemuan ini menunjukkan bahawa data sintetik adalah alat berkesan bagi meningkatkan klasifikasi EEG dalam kajian kesan pendedahan RF.

Downloads

Download data is not yet available.

Metrics

Metrics Loading ...

References

Liang Y, Jiang J, Chen Y, Zhu R, Lu C, Wang Z (2019). Optimized feedforward neural network training for efficient Brillouin frequency shift retrieval in fiber. IEEE Access, 7: 68034–68042. https://doi.org/10.1109/ACCESS.2019.2919138

Ojha VK, Abraham A, Snasel V (2017). Metaheuristic design of feedforward neural networks: A review of two decades of research. Eng Appl Artif Intell, 60: 97–116. https://doi.org/10.1016/j.engappai.2017.01.013

Nemzer LR, Cravens GD, Worth RM, Motta F, Placzek A, Castro V, Lou JQ (2021). Critical and ictal phases in simulated EEG signals on a small-world network. Frontiers in Computational Neuroscience, 14: 1–13. https://doi.org/10.3389/fncom.2020.583350

Xu Y, Yang J, Sawan M (2022). Multichannel synthetic preictal EEG signals to enhance the prediction of epileptic seizures. IEEE Transactions on Biomedical Engineering, 69(12): 3572–3583. https://doi.org/10.1109/TBME.2022.3171982

Leon J, Escobar JJ, Ortiz A, Ortega J, González J, Martín-Smith P, Gan JQ (2020). Deep learning for EEG-based motor imagery classification: Accuracy–cost trade-off. PLoS One, 15(6): 1–23. https://doi.org/10.1371/journal.pone.0234178

Ding W, Liu A, Guan L, Chen X (2024). A novel data augmentation approach using mask encoding for deep learning-based asynchronous SSVEP-BCI. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 32: 875–886. https://doi.org/10.1109/TNSRE.2024.3366930

Cekic M (2023). Anomaly detection in medical time series with generative adversarial networks: A selective review. In: Anomaly Detection – Recent Advances, AI and ML Perspectives and Applications. IntechOpen, 1–18. https://doi.org/10.5772/intechopen.112582

Carrle FP, Hollenbenders Y, Reichenbach A (2023). Generation of synthetic EEG data for training algorithms supporting the diagnosis of major depressive disorder. Frontiers in Neuroscience, 17: 1–11. https://doi.org/10.3389/fnins.2023.1219133

Habashi AG, Azab AM, Eldawlatly S, Aly GM (2023). Generative adversarial networks in EEG analysis: An overview. Journal of NeuroEngineering and Rehabilitation, 20(1): 1–16. https://doi.org/10.1186/s12984-023-01169-w

Yadav P, Gaur M, Madhukar RK, Verma G, Kumar P, Fatima N, Sarwar S, Dwivedi YR (2024). Rigorous experimental analysis of tabular data generated using TVAE and CTGAN. International Journal of Advanced Computer Science & Applications, 15(4): 1250–1262. https://doi.org/10.14569/IJACSA.2024.01504125

Xu L, Skoularidou M, Cuesta-Infante A, Veeramachaneni K (2019). Modeling tabular data using conditional GAN. Advances in Neural Information Processing Systems, 32: 7333–7343. https://doi.org/10.48550/arXiv.1907.00503

Yang Q, Yan P, Zhang Y, Yu H, Shi Y, Mou X, Kalra MK, Zhang Y, Sun L, Wang G (2018). Low-dose CT image denoising using a generative adversarial network with Wasserstein distance and perceptual loss. IEEE Transactions on Medical Imaging, 37(6): 1348–1357. https://doi.org/10.1109/TMI.2018.2827462

Zietsch BP, Hansen JL, Hansell NK, Geffen GM, Martin NG, Wright MJ (2007). Common and specific genetic influences on EEG power bands delta, theta, alpha, and beta. Biol Psychol, 75(2): 154–164. https://doi.org/10.1016/j.biopsycho.2007.01.004

Michel CM, Lehmann D, Henggeler B, Brandeis D (1992). Localization of the sources of EEG delta, theta, alpha and beta frequency bands using the FFT dipole approximation. Electroencephalogr Clin Neurophysiol, 82: 38–44. https://doi.org/10.1016/0013-4694(92)90180-P

Pant A, Kumar A, Verma C, Illés Z (2024). Comparative exploration on EEG signal filtering using window control methods. Results in Control and Optimization, 17: 1–16. https://doi.org/10.1016/j.rico.2024.100485

Ahmed MZI, Sinha N, Phadikar S, Ghaderpour E (2022). Automated feature extraction on AsMap for emotion classification using EEG. Sensors, 22(6): 1–17. https://doi.org/10.3390/s22062346

Chikhi S, Matton N, Blanchet S (2022). EEG power spectral measures of cognitive workload: A meta-analysis. Psychophysiology, 59(6): 1–16. https://doi.org/10.1111/psyp.14009

Ismail LE, Karwowski W (2020). Applications of EEG indices for the quantification of human cognitive performance: A systematic review and bibliometric analysis. PLoS One, 15(12): 1–29. https://doi.org/10.1371/journal.pone.0242857

Ambarwari SA, Adrian QJ, Herdiyeni Y (2020). Analysis of the effect of data scaling on the performance of the machine learning algorithm for plant identification. Jurnal RESTI, 4(1): 117–122. https://doi.org/10.29207/resti.v4i1.1565

Ahmed HA, Muhammad Ali PJ, Faeq AK, Abdullah SM (2022). An investigation on disparity responds of machine learning algorithms to data normalization method. Aro-The Scientific Journal of Koya University, 10(2): 29–37. https://doi.org/10.14500/aro.10970

Singh D, Singh B (2020). Investigating the impact of data normalization on classification performance. Applied Soft Computing, 97: 1–15. https://doi.org/10.1016/j.asoc.2020.105524

Gunturkun O, Strockens F, Ocklenburg S (2020). Brain lateralization: A comparative perspective. Physiol Rev, 100(3): 1019–1063. https://doi.org/10.1152/physrev.00006.2019

Rogers LJ (2021). Brain lateralization and cognitive capacity. Animals, 11(7): 1–11. https://doi.org/10.3390/ani11071996

Szeghalmy S, Fazekas A (2023). A comparative study of the use of stratified cross-validation and distribution-balanced stratified cross-validation in imbalanced learning. Sensors, 23(4): 1–16. https://doi.org/10.3390/s23042333

Liu Q, Li C, Wanga V, Shepherd BE (2018). Covariate-adjusted Spearman’s rank correlation with probability-scale residuals. Biometrics, 74(2): 595–605. https://doi.org/10.1111/biom.12812

Arvanitis TN, White S, Harrison S, Chaplin R, Despotou G (2022). A method for machine learning generation of realistic synthetic datasets for validating healthcare applications. Health Informatics Journal, 28(2): 1–13. https://doi.org/10.1177/14604582221077000

Li Z, Zhu H, Lu Z, Yin M (2023). Synthetic data generation with large language models for text classification: Potential and limitations. arXiv preprint, 1–15. Retrieved from http://arxiv.org/abs/2310.07849

Namazi H, Aghasian E, Ala TS (2020). Complexity-based classification of EEG signal in normal subjects and patients with epilepsy. Technology and Health Care, 28(1): 57–66. https://doi.org/10.3233/THC-191979

You Y, Li Y, Yu B, Ying A, Zhou H, Zuo G, Xu J (2024). A study on EEG differences between active counting and focused breathing tasks for more sensitive detection of consciousness. Frontiers in Neuroscience, 18: 1–14. https://doi.org/10.3389/fnins.2024.1341986

Isa RM, Pasya I, Taib MN, Jahidin AH, Omar WRW, Fuad N (2012). Assessment of brainwave asymmetry and hemisphere dominance due to RF radiation. Proceedings – UKSim-AMSS 6th European Modelling Symposium, 153–157. https://doi.org/10.1109/EMS.2012.89

Ren B, Pan Y (2023). Extracting and supplementing method for EEG signal in manufacturing workshop based on deep learning of time–frequency correlation. J Intell Manuf, 34(7): 3179–3196. https://doi.org/10.1007/s10845-022-01997-y

Massar H, Belhoussine Drissi T, Nsiri B, Miyara M (2025). Advancements in blind source separation for EEG artifact removal: A comparative analysis of variational mode decomposition and discrete wavelet transform approaches. Applied Acoustics, 228: 1–12. https://doi.org/10.1016/j.apacoust.2024.110300