A New method for Improvement of the Accuracy of Character Recognition in P300 Speller System: Optimization of Channel Selection by Using Recursive Channel Elimination Algorithm Based on Deep Learning

Document Type : Original Article

Authors

1 Department of Electrical Engineering and Information Technology, Iranian Research Organization for Science and Technology, Tehran, Iran

2 Faculty of Electrical, Biomedical and Mechatronics Engineering, Qazvin Branch, Islamic Azad University, Qazvin, Iran

Abstract

Background and purpose:P300 speller is a kind of Brain-Computer Interface (BCI) system in which the user may type words by using the responses obtained from human focus on different characters. The high sensitivity of brain signals against noise in parallel with the similarity of responses obtained from the user focus on different characters makes it difficult to classify the characters based on their respective P300 wave. On the other hand, all areas of the brain does not carry useful P300 information.
Proposed Methods: In this study, a new method is proposed to improve the performance of speller system which is based on selecting optimal P300 channels. In the proposed method, recursive elimination algorithm is presented for channel optimization, which utilizes deep learning concept (e.g. Convolutional Neural Network) as its cost function. The proposed method is examined on a data set from EEG signals recorded in a P300 speller system, including 64 different channels of responses to 29 characters. Then, its performance is compared with some existing methods.
Results: The obtained results showed the ability of the proposed method in recognizing the characters in such way that it could accurately (i.e. 97.34%) detect 29 characters by using only 24 out of all 64 electrodes.
Conclusion: Applying the proposed method in speller systems led to considerable improvement in classification of characters compared to its alternatives. Several experiments proved that utilizing the proposed scheme may increases the accuracy almost 12.9 percent compared to non-optimized case in parallel with reduction of the number of involved channels by approximately 1/3. Based on these results, the proposed method may be considered as an effective choice for application in P300 speller systems, thanks to reduction of the complexity of the system which is caused by the  reduced number of channels and, on the other hand, due to its potential in increasing the accuracy of character recognition. 
 

Keywords

References

1. Hassanien AE, Azar A. Brain-computer interfaces. Switzerland: Springer. 2015. doi: 10.1007/978-3-319-10978-7 .
2. Bansal D, Mahajan R. EEG-Based Brain-Computer Interfaces: Cognitive Analysis and Control Applications. Germany: Academic Press; 2019..
3. Colwell KA, Ryan DB, Throckmorton CS, Sellers EW, Collins LM. Channel selection methods for the P300 Speller. J Neurosci Methods. 2014;232:6-15. doi: 10.1016/j.jneumeth.2014.04.009.
4. Sanei S, Chambers J. EEG Signal Processing, Centre of Digital Signal Processing, Cardiff University, UK. Publicação acadêmica, 1ª Edição, Elsevier. 2007;16.
5. Magee JC, Winhusen T. The coupling of nicotine and stimulant craving during treatment for stimulant dependence. J Consult Clin Psychol. 2016;84(3):230-7. doi: 10.1037/ccp0000054.
6. Shahriari Y, Erfanian A, editors. A mutual information based channel selection scheme for P300-based brain computer interface. 2011 5th International IEEE/EMBS Conference on Neural Engineering; 2011: p. 434–7.
7. Rosas-Cholula G, Ramirez-Cortes JM, Alarcon-Aquino V, Martinez-Carballido J, Gomez-Gil P, editors. On signal P-300 detection for BCI applications based on wavelet analysis and ICA preprocessing. 2010 IEEE Electronics, Robotics and Automotive Mechanics Conference; 2010: IEEE. 10.1109/cerma.2010.48.
8. Selim AE, Wahed MA, Kadah YM, editors. Machine learning methodologies in P300 speller Brain-Computer Interface systems. 2009 National Radio Science Conference; 2009: IEEE.
9. Manyakov NV, Chumerin N, Combaz A, Van Hulle MM. Comparison of classification methods for P300 brain-computer interface
on disabled subjects. Comput Intell Neurosci. 2011;2011:519868. doi: 10.1155/2011/519868.
10. Santamaria-Vazquez E, Martinez-Cagigal V, Gomez-Pilar J, Hornero R. Asynchronous Control of ERP-Based BCI Spellers Using
Steady-State Visual Evoked Potentials Elicited by Peripheral Stimuli. IEEE Trans Neural Syst Rehabil Eng. 2019;27(9):1883-92. doi: 10.1109/TNSRE.2019.2934645.
11. Kübler A, Botrel L, editors. Imagining the P300 Speller: Good idea or nonsense? 2019 7th International Winter Conference on Brain- Computer Interface (BCI); 2019.
12. Gu Z, Chen Z, Zhang J, Zhang X, Yu ZL. An Online Interactive Paradigm for P300 Brain- Computer Interface Speller. IEEE Trans Neural Syst Rehabil Eng. 2019;27(2):152-61. doi: 10.1109/TNSRE.2019.2892967.
13. Kaper M, Meinicke P, Grossekathoefer U, Lingner T, Ritter H. BCI Competition 2003-Data set IIb: support vector machines for the P300 speller paradigm. IEEE Trans Biomed Eng. 2004;51(6):1073-6. doi: 10.1109/TBME.2004.826698.
14. Rakotomamonjy A, Guigue V. BCI competition III: dataset II- ensemble of SVMs for BCI P300 speller. IEEE Trans Biomed Eng. 2008;55(3):1147-54. doi: 10.1109/TBME.2008.915728.
15. Hoffmann U, Garcia G, Vesin J-M, Diserens K, Ebrahimi T, editors. A boosting approach to P300 detection with application to braincomputer interfaces. Conference Proceedings 2nd International IEEE EMBS Conference on Neural Engineering, 2005; 2005: IEEE. doi: 10.1109/cne.2005.1419562.
16. Liang N, Bougrain L, editors. Averaging techniques for single-trial analysis of oddball event-related potentials2008.
17. Mirghasemi H, Fazel-Rezai R, Shamsollahi MB. Analysis of p300 classifiers in brain computer interface speller. Conf Proc IEEE Eng Med Biol Soc. 2006;2006:6205-8. doi: 10.1109/IEMBS.2006.259521.
18. Ang KK, Chin ZY, Zhang H, Guan C. Mutual information-based selection of optimal spatial–temporal patterns for single-trial EEG-based BCIs. Pattern Recognition. 2012;45(6):2137-44. doi: 10.1016/j.patcog.2011.04.018.
19. Cecotti H, Rivet B, Congedo M, Jutten C, Bertrand O, Maby E, et al. A robust sensorselection method for P300 brain-computer
interfaces. J Neural Eng. 2011;8(1):016001. doi: 10.1088/1741-2560/8/1/016001.
20. Shan H, Liu Y, Stefanov TP, editors. A Simple Convolutional Neural Network for Accurate P300 Detection and Character Spelling in Brain Computer Interface. IJCAI; 2018.doi: 10.24963/ijcai.2018/222.
21. Zhang X, Yao L, Wang X, Monaghan J, Mcalpine D, Zhang Y. A Survey on Deep Learning based Brain Computer Interface: Recent Advances and New Frontiers. arXiv 2019. arXiv preprint arXiv:190504149.
22. Kundu S, Ari S. MsCNN: A Deep Learning Framework for P300 Based Brain-Computer Interface Speller. IEEE Transactions on Medical Robotics and Bionics. 2019.
23. American Electroencephalographic Society guidelines for standard electrode position nomenclature. J Clin Neurophysiol. 1991;8(2):200-2.
24. Kee C-Y, Ponnambalam SG, Loo C-K. Multiobjective genetic algorithm as channel selection method for P300 and motor imagery data set. Neurocomputing. 2015;161:120-31. doi: 10.1016/j.neucom.2015.02.057.
25. Xu M, Qi H, Ma L, Sun C, Zhang L, Wan B, et al. Channel selection based on phase measurement in P300-based brain-computer interface. PLoS One. 2013;8(4):e60608. doi: 10.1371/journal. pone.0060608.
26. Krusienski D, Schalk G. Wadsworth BCI dataset (P300 evoked potentials). BCI Competition III Challenge. 2004.
27. Cecotti H, Graser A. Convolutional neural networks for P300 detection with application to brain-computer interfaces. IEEE Trans Pattern Anal Mach Intell. 2011;33(3):433-45. doi: 10.1109/TPAMI.2010.125.
28. Hillyard SA, Kutas M. Electrophysiology of cognitive processing. Annu Rev Psychol. 1983;34:33-61. doi: 10.1146/annurev.ps.34.020183.000341.
Health Management & Information Science
Volume 7, Issue 1
January 2020
Pages 42-51

Files

Share

How to cite

Statistics