A Bayesian Tri-Level Optimization Framework for Optimal EEG Channel Selection in Seizure Prediction

Authors
Maria Afsharirad
Maryam Khansari
Abstract
In this study, a tri-level optimization framework is proposed for optimal EEG channel selection with the aim of improving seizure prediction accuracy. This approach is grounded in Bayesian theory and developed based on principles of tri-level optimization methods. In the first level, the conditional probability derived from the correlation between temporal segments of preictal and ictal EEG signals is calculated. By incorporating the prior probability, the posterior probability is modeled and optimized as the objective function of the first level, which is applied, at the second level, to determine the optimal seizure prediction time for each channel, representing the earliest moment when seizure-related information becomes evident in that channel. The third-level optimization problem seeks to identify the optimal EEG channel using the optimal timing of each channel. This tri-level process ensures the selection of a channel that not only minimizes prediction delay but also provides the most informative signal for the system. The implementation of this method on real data from 14 patients with epilepsy demonstrates that the proposed approach can achieve an efficient trade-off between prediction accuracy and model complexity.
Keywords
Bayesian model
multi-level optimization
channel selection
Correlation analysis
EEG signal
probabilistic modeling
[1] Moctezuma, L. A., & Molinas, M. (2020). EEG channel-selection method for epileptic-seizure classification based on multi-objective optimization. Frontiers in neuroscience, 14, 593.

[2] R. Jana and I. Mukherjee, "Efficient seizure prediction and EEG channel selection based on multi-objective optimization," IEEE Access, vol. 11, pp. 54112–54121, 2023.

[3] J. S. Ra, T. Li, and Y. Li, "A novel permutation entropy-based EEG channel selection for improving epileptic seizure prediction," Sensors, vol. 21, no. 23, p. 7972, 2021.

[4] T. Hastie, R. Tibshirani, and J. Friedman, The Elements of Statistical Learning. New York, NY, USA: Springer, 2009.

[5] R. Cooper, J. W. Osselton, and J. C. Shaw, EEG Technology. Oxford, UK: Butterworth-Heinemann, 2014.

[6] K. Blinowska and P. Durka, "Electroencephalography (EEG)," in Wiley Encyclopedia of Biomedical Engineering, vol. 10, John Wiley & Sons, 2006, p. 9780471740360.

[7] P. Detti, "Siena scalp EEG database," PhysioNet, vol. 10, p. 493, 2020.

[8] T. Dissanayake, T. Fernando, S. Denman, S. Sridharan, and C. Fookes, "Geometric deep learning for subject independent epileptic seizure prediction using scalp EEG signals," IEEE Journal of Biomedical and Health Informatics, vol. 26, no. 2, pp. 527-538, 2021.

[9] P. Bourke, "Cross correlation," in Cross Correlation, Auto Correlation—2D Pattern Identification, 1996, p. 596.

[10] H. Akoglu, "User’s guide to correlation coefficients," Turkish Journal of Emergency Medicine, vol. 18, no. 3, pp. 91–93, 2018, doi: 10.1016/j.tjem.2018.08.001.

[11] P. Schober, C. Boer, and L. A. Schwarte, “Correlation coefficients: Appropriate use and interpretation,” Anesthesia & Analgesia, vol. 126, no. 5, pp. 1763–1768, May 2018, doi: 10.1213/ANE.0000000000002864.

[12] ایران., رستگار، س.، پولادیان، م.، نیکخو، م. (۱۴۰۱). تعیین همبستگی بین دو سیگنال EMG و EEG ثبت‌شده به‌صورت هم‌زمان در قدرت انقباض‌های مختلف چنگ زدن دست. پنجمین همایش ملی فناوری‌های نوین در مهندسی برق، کامپیوتر و مکانیک

[13] J. C. Yoo and T. H. Han, “Fast normalized cross-correlation,” Circuits, Systems and Signal Processing, vol. 28, no. 6, pp. 819–843, 2009, doi: 10.1007/s00034-009-9130-7

[14] J. Mockus, "The application of Bayesian methods for seeking the extremum," Towards Global Optimization, vol. 2, pp. 117, 1998.

[15] M. E. Glickman and D. A. Van Dyk, "Basic Bayesian methods," Topics in Biostatistics, pp. 319-338, 2007.

[16] Williams, C. K., & Rasmussen, C. E. (2006). Gaussian processes for machine learning (Vol. 2, No. 3, p. 4). Cambridge, MA: MIT press.

[17] S. Ryu and I. Joe, “A Hybrid DenseNet-LSTM Model for Epileptic Seizure Prediction,” Applied Sciences, vol. 11, no. 16, Art. no. 7661, Aug. 2021, doi: 10.3390/app11167661.

[18] Moctezuma, L. A., & Molinas, M. (2020). EEG channel-selection method for epileptic-seizure classification based on multi-objective optimization. Frontiers in neuroscience, 14, 593.‌

[19] F. E. Polat, S. Güney, and A. K. Cömert, "Synchronization-based EEG channel selection and seizure prediction using Siena Scalp EEG database," Processes, vol. 8, no. 7, p. 846, 2020.