RECURRENT NEURAL NETWORK BASED T WAVE END DETECTION

Year 2020, Volume: 9 Issue: 1, 622 - 636, 30.01.2020

Mehmet İşcan Aydın Yeşildirek

https://doi.org/10.28948/ngumuh.681169

Abstract

Artificial neural networks (ANN) have been utilized in many areas such as classification, pattern discrimination, etc. In this study, a novel ANN based algorithm is proposed to detect T-wave end point, one of the important ECG signal parameters to determine heart health conditions that may be hardly marked with respect to the other ECG fiducial points. Since the feedforward neural networks (NN) applications in time series signal processing has a limited success, we adopt a recurrent NN architecture. To demonstrate the effectiveness of the algorithm, a set of single channel ECG signals obtained from PHYSIONET/QT database is preprocessed first, then using some clustering algorithms, the observation of the T wave end is sought within a limited time window. The algorithm is shown to reach a performance exceeding the expectations of the standards. It has been trained by 55 beats out of 295 and reaches the success rate of 11.16±6.16 and -4.70±6.64 milliseconds error in terms of absolute and non-absolute conditions, respectively. Furthermore, the proposed algorithm has been retested on some other 421 ECG beats annotated by experts, where it had an excellent score of -6.40±17.22 milliseconds marking error.

Keywords

Recurrent neural network, T Wave end, Signal processing, Classification

GERİ DÖNÜŞÜMLÜ YAPAY SİNİR AĞI TABANLI T DALGASI SONU TESPİTİ

Abstract

Günümüzde yapay sinir ağlarının (YSA) kullanımı sınıflama işlemlerinden özel kalıplar tanımaya kadar birçok yerde karşımıza çıkmaktadır. Bu çalışmada, YSA tabanlı özgün bir algoritma ile kalp sağlığının önemli göstergelerinden biri olup, işaretlenmesi diğer EKG referans noktalarına göre güç olan T dalga sonunun tespiti önerilmiştir. Genel olarak, zamana bağlı sinyal serilerinin işlenmesinde ileri sinir ağları sınırlı başarı sağlanabildiğinden geri dönüşümlü YSA mimarisi kullanılmıştır. Önerilen algoritmanın etkinliğinin gösterimi için, PHYSIONET/QT veri tabanında bulunan tek kanaldan elde edilen EKG sinyalleri önişlemlere geçirilerek, kümeleme diyagramları ile T dalgasının son noktası sıkıştırılmış bir alana indirgenmiştir. Algoritmadan elde edilen sonuçlar, bu alanda standartlarda ifade edilen beklentilerin üzerinde bir performans sergilemiştir. 55’i eğitim atımı olan toplam 295 atımda mutlak olan hata değerlerinde 11.16±6.16 milisaniye, mutlak olmayan hata değerlerinde ise -4.70±6.64 milisaniyelik iyi bir performansa ulaşılmıştır. Ayrıca geliştirilen bu yöntem, önceden eğitilmemiş ve uzmanlar tarafından işaretlenmiş 421 yeni atım üzerinde denendiğinde, -6.40±17.22 milisaniye gibi çok iyi bir mutlak olmayan işaretleme hata değerine ulaşılmıştır.

Keywords

Geri-Dönüşümlü sinir ağları, T dalgası sonu, Sinyal İşleme, Sınıflama

References

• [1] R. J. Goldberg, J. Bengtson, Z. Chen, K. M., Anderson, E. Locati, D.Levy, "Duration of the QT interval and total and cardiovascular mortality in healthy persons (The Framingham Heart Study experience),” The American journal of cardiology, vol. 67, no. 1, pp. 55-58.,1991.
• [2] J. P. Martínez, R. Almeida, S. Olmos, A. P. Rocha, P. Laguna "A wavelet-based ECG delineator: evaluation on standard databases,” IEEE Transactions on biomedical engineering, vol. 51, no. 4, pp. 570-581, 2004.
• [3] M. Işcan, Ş. Sariozkan, A. Yilmaz, & C. Yilmaz, “Multilead QT interval analysis algorithm based on continuous wavelet transform,” In 4th International Conference on IEEE in Electrical and Electronic Engineering (ICEEE)’ 05, 2017 pp. 318-322.
• [4] J. García, L. Sornmo, S. Olmos, P. Laguna “Automatic detection of ST-T complex changes on the ECG using filtered RMS difference series: application to ambulatory ischemia monitoring,” IEEE Transactions on Biomedical Engineering, vol. 47, no. 9, pp. 1195-1201, 2000.
• [5] I. K. Daskalov & I. I. Christov, “Automatic detection of the electrocardiogram T-wave end,” Medical and Biological Engineering and Computing, vol. 37, np. 3, pp. 348-353, 1999.
• [6] Q. Zhang, A. I. Manriquez, C. Médigue, Y. Papelier, M. Sorine, “An algorithm for robust and efficient location of T-wave ends in electrocardiograms,” IEEE Transactions on Biomedical Engineering, vol. 53, no. 12, pp. 2544-2552, 2006.
• [7] P. Laguna, N. V. Thakor, P. Caminal, R. Jane, H. R. Yoon, A. Bayés de Luna & J. Guindo, "New algorithm for QT interval analysis in 24-hour Holter ECG: performance and applications,” Medical and Biological Engineering and Computing, vol. 28, no. 1, pp. 67-73, 1990.
• [8] Y. C. Chesnokov, D. Nerukh, R. C. Glen, “Individually adaptable automatic QT detector,” In IEEE Computers in Cardiology, 2006, pp. 337- 340.
• [9] C. Yılmaz, M. İscan, A. A. Yılmaz, “A Fully Automatic Novel Method to Determine QT Interval Based on Continuous Wavelet Transform,” Journal Journal of Electrical & Electronics Engineering, vol. 17, no. 1, pp. 3103-3111, 2017.
• [10] M. İşcan, C. Yilmaz, F. Yiğit, "T-wave end pattern classification based on Gaussian mixture model,” In 24th IEEE Signal Processing and Communication Application Conference (SIU), 2016, pp. 1953 – 1956.
• [11] M. İşcan &C. Yılmaz, “QT Zaman Aralığının Gauss Karışım Modeli ve Yapay Sinir Ağı Tabanlı Tespiti,” Ömer Halisdemir Üniversitesi Mühendislik Bilimleri Dergisi, vol. 6, no. 2, pp. 752-762, 2017
• [12] D. Hayn, A. Kollmann, G. Schreier, “Automated QT interval measurement from multilead ECG signals,” In IEEE Computers in Cardiology, 2006, pp. 381-384.
• [13] G. D. Clifford, M. C. Villarroel, “Model-based determination of QT intervals,” In IEEE Computers in Cardiology, 2006, pp. 357-360.
• [14] M. M. Gupta, L. Jın, & N. Homma, Static and Dynamic Neural Networks from Fundamentals to Advanced Theory. New Jersey : A John Wiley & Sıns Inc, 2003, pp. 106.
• [15] E. Alpaydin, Introduction to machine learning. USA-Cambridge : MIT press, 2014.
• [16] J.M. Lindauer, R.E. Gregg, E.D. Helfenbein, M. Shao, S.H. Zhou, "Global QT measurements in the Philips 12-lead algorithm,” Journal of Electrocardiology, vol. 38, no. 1, pp. 90, 2005.
• [17] D. Romero, J. P. Martínez, P. Laguna, E. Pueyo “Ischemia detection from morphological QRS angle changes,” Journal of Physiological measurement, vol. 37, no. 7, pp. 1004, 2016.
• [18] CSE Working Party. “Recommendations for measurement standards in quantitative electrocardiography,” Journal of European Heart Journal, vol. 6, no. 10, pp. 815-825, 1985.