Transformer Kodlayıcı ve Zaman-Frekans Görüntüleri Kullanarak Otomatik Uyku Evreleri Sınıflandırması

Year 2023, , 103 - 108, 18.10.2023

Göksu Zekiye Özen , Yunus Özen

https://doi.org/10.53070/bbd.1370639

Abstract

Bu çalışmada Polisomnografi (PSG) kayıtlarından alınan tek kanallı EEG verileri kullanarak otomatik uyku evreleri sınıflandırması yapan bir derin öğrenme modeli önerilmektedir. Önerilen model, EEG sinyallerinin kısa süreli Fourier dönüşümü (STFT) ile elde edilen zaman-frekans görüntülerinden öznitelik çıkarmak için Transformer kodlayıcı kullanmaktadır. Transformer kodlayıcının çok başlı dikkat mekanizması, zaman-frekans görüntülerindeki zaman bağımlılıklarını yakalayarak modelin uykunun sıralı doğasını anlama performansını artırmaktadır. Önerilen modelin performansı, SleepEDF Expanded adlı açık erişim veri seti üzerinde değerlendirilmiştir ve 0.84 F1 skoru ile yüksek doğruluk değerine sahip sonuç elde edilmiştir. Modelin zaman-frekans görüntüleri kullanması, EEG sinyallerinin temel zaman alanı ve frekans alanı özelliklerini yakalayarak doğru uyku evreleri sınıflandırmasına katkı sağlamaktadır. Gelecek çalışmalarda, diğer PSG kanalları da dâhil edilerek uygulamada kullanımı mümkün olabilecek bir model geliştirilebileceği değerlendirilmektedir.

Keywords

Uyku evreleri sınıflandırması, Derin öğrenme, Transformer kodlayıcılar, EEG sinyalleri, Polisomnografi, Zaman-frekans görüntüleri

Automated Sleep Stage Classification Using Transformer Encoders and Time-Frequency Images

Abstract

This study proposes a deep-learning model for automatic sleep stage classification using single-channel EEG data from polysomnography (PSG) recordings. The model employs transformer encoders to extract features from time-frequency images obtained through short-time Fourier transform (STFT) of the EEG signals. The transformer encoder's multi-head attention mechanism captures temporal dependencies within the time-frequency images, improving the model's ability to understand the sequential nature of sleep. We evaluated the model's performance on the publicly available SleepEDF Expanded dataset, and a high accuracy of 0.84 F1 score was obtained. The model's use of time-frequency images enables it to capture essential time-domain and frequency-domain features of EEG signals, contributing to accurate sleep stage classification. In conclusion, our deep learning model based on transformer encoders provides an efficient and reliable solution for sleep stage classification from single-channel EEG data. Future research may explore extending the model to incorporate additional PSG channels and expanding its utility to broader sleep studies.
Keywords

Keywords

Sleep stage classification, Deep learning, Transformer encoders, EEG signals, Polysomnography, Time-frequency images

References

• Chen, P. C., Zhang, J., Thayer, J. F., & Mednick, S. C. (2022). Understanding the roles of central and autonomic activity during sleep in the improvement of working memory and episodic memory. Proceedings of the National Academy of Sciences, 119(44).
• Chriskos, P., Frantzidis, C. A., Nday, C. M., Gkivogkli, P. T., Bamidis, P. D., & Kourtidou-Papadeli, C. (2021). A review on current trends in automatic sleep staging through bio-signal recordings and future challenges. Sleep Medicine Reviews, 55.
• Figueiro, M. G., & Pedler, D. (2023). Cardiovascular disease and lifestyle choices: Spotlight on circadian rhythms and sleep. Progress in Cardiovascular Diseases.
• Fiorillo, L., Puiatti, A., Papandrea, M., Ratti, P. L., Favaro, P., Roth, C., ... & Faraci, F. D. (2019). Automated sleep scoring: A review of the latest approaches. Sleep Medicine Reviews, 48, 101204.
• Goldberger, A. L., Amaral, L. A. N., Glass, L., Hausdorff, J. M., Ivanov, P. C., Mark, R. G., Mietus, J. E., Moody, G. B., Peng, C. K., & Stanley, H. E. (2000). PhysioBank, PhysioToolkit, and PhysioNet - Components of a new research resource for complex physiologic signals. Circulation, 101(23): E215-E220.
• Hand, D. J., Till, R. J, (2001). A Simple Generalisation of the Area Under the ROC Curve for Multiple Class Classification Problems. Machine Learning, 45(2): 171-186.
• Iber, C. (2007). The AASM manual for the scoring of sleep and associated events: Rules, Terminology and Technical Specifications.
• Malekzadeh, M., Hajibabaee, P., Heidari, M., & Berlin, B. (2022, January). Review of deep learning methods for automated sleep staging. In 2022 IEEE 12th Annual Computing and Communication Workshop and Conference (CCWC), pp. 80-86.
• Ozen, G. Z., SULTANOV, R., Ozen Y., & Gunes, Z. Y. (2020). A Convolutional Neural Network Based on Raw Single Channel EEG for Automatic Sleep Staging. Sakarya University Journal of Computer and Information Sciences 3(2): 149-158.
• Rundo, J. V., & Downey III, R. (2019). Polysomnography. Handbook of Clinical Neurology, 160: 381-392.
• Sekkal, R. N., Bereksi-Reguig, F., Ruiz-Fernandez, D., Dib, N., & Sekkal, S. (2022). Automatic sleep stage classification: From classical machine learning methods to deep learning. Biomedical Signal Processing and Control, 77:
• Sri, T. R., Madala, J., Duddukuru, S. L., Reddipalli, R., & Polasi, P. K. (2022, Nisan). A Systematic Review on Deep Learning Models for Sleep Stage Classification. In 2022 6th International Conference on Trends in Electronics and Informatics (ICOEI), pp. 1505-1511.
• Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, Ł., & Polosukhin, I. (2017). Attention is all you need. In Proceedings of the 31st International Conference on Neural Information Processing Systems, Long Beach, California, USA, pp. 6000–6010.
• Wolpert, E. A. (1969). A manual of standardized terminology, techniques and scoring system for sleep stages of human subjects. Electroencephalography & Clinical Neurophysiology, 26(2): 644-644