心律不整是指心臟電傳導系統異常所引起的各種症狀，一般來說是透過觀測心電圖來進行診斷，傳統的心臟疾病自動檢測往往使用PQRST等波型之特徵點進行心律不整之判別，但此種方式及易受到雜訊之干擾。隨著深度學習技術方面的成熟，許多研究也將其導入心臟疾病判別中，並取得顯著的成效。
本文開發了一種有效的深度學習架構能同時達到去雜訊且提升辦別準確度的成效。在心律不整自動化偵測系統的實際應用場景中，往往會因為不同的資料庫或是錄製儀器，需要再藉由人為的經驗使用不同的前處理方式才能獲得更穩健的結果，為了克服這點，本文提出一種新的心臟疾病檢測架構: 藉由導入時頻相關注意力機制，讓網路能自行學習該如何對訊號進行前處理。同時，將多尺度的洞卷積架構導入網路中，讓其能在不降低維度的情況下提取不同尺度的訊號特徵，最終結合頻譜圖與一維訊號的網路結果以判斷最後的心律不整種類。在定性與定量實驗結果裡，我們的方法皆優於過往技術，且在資料庫中，我們也可以得到更穩健的訊號特徵及更好的辨識結果。

Arrhythmia refers to various symptoms caused by abnormal electrical conduction system of the heart, and is generally diagnosed by observing an electrocardiogram. Traditionally, arrhythmia is often discriminated by using the fiducial points of the electrocardiogram waveform such as PQRST, which is susceptible to interference from noise. As the maturity of deep learning technology nowadays, numbers of study start to introduce it into arrhythmia detection, and achieve significant result.
In this thesis, a cardiac arrhythmia diagnostic network based on the attention deep learning architecture is proposed. To overcome the issue introduced from signal preprocessing variety, which is normally caused by different record equipment and situation, the proposed network architecture adopts the time-frequency fusion module that considers the correlation between the original signal and spectrogram to facilitate the gathering of the more reliable features on arrhythmia detection. Moreover, multiple scales of atrous convolutions are applied in the proposed network, enabling the network to capture more stable features and achieve better result.
As shown in the experimental results, the proposed network achieves significantly superior result comparing with former methods. This study confirmed that the proposed algorithm can be a very competitive candidate for cardiac arrhythmia diagnosis.

[1] https://zh.wikipedia.org/wiki/%E5%BF%83%E7%94%B5%E5%9B%BE
[2] Pinto, João Ribeiro, Jaime S. Cardoso, and André Lourenço. "Evolution, current challenges, and future possibilities in ECG biometrics." IEEE Access 6 (2018): 34746-34776.
[3] F. Agrafioti and D. Hatzinakos. pp. 82–87, ‘‘ECG based recognition using second order statistics,’’ in Proc. 6th Annu. Commun. Netw. Services Res. Conf. (CNSR), May 2008
[4] A. Sahayadhas, K. Sundaraj, and M. Murugappan. ‘‘Detecting driver drowsiness based on sensors: A review,’’ Sensors, vol. 12, no. 12, pp. 16937–16953, 2012
[5] Hannun, Awni Y., et al. "Cardiologist-level arrhythmia detection and classification in ambulatory electrocardiograms using a deep neural network." Nature medicine 25.1 (2019): 65.
[6] R. Hoekema, G. J. H. Uijen, and A. van Oosterom. ‘‘Geometrical aspects of the inter-individual variability of multilead ECG recordings,’’ in Proc. Comput. Cardiology, vol. 26, Sep 1999, pp. 499–502,
[7] A. van Oosterom, R. Hoekema, and G. J. H. Uijen, ‘‘Geometrical factors affecting the interindividual variability of the ECG and the VCG,’’ J. Electrocardiol., vol. 33, 2000, pp. 219–227
[8] R. Hoekema, G. J. H. Uijen, and A. van Oostero. ‘‘Geometrical aspects of the interindividual variability of multilead ECG recordings,’’ IEEE Trans. Biomed. Eng., vol. 48, no. 5, pp. 551–559. May 2001.
[9] R. J. A. Schijvenaars, ‘‘Intra-individual Variability of the Electrocardiogram: Assessment and exploitation in computerized ECG analysis.’’ Ph.D. dissertation, Dept. Med. Inform., Erasmus Univ. Rotterdam, Rotterdam, The Netherlands. 2000.
[10] F. Agrafioti, F. M. Bui, and D. Hatzinakos, ‘‘Secure telemedicine: Biometrics for remote and continuous patient verification,’’ J. Comput. Netw. Commun., vol. Art. no. 924791. 2012 Oct.
[11] M. Kyoso and A. Uchiyama, ‘‘Development of an ECG identification system,’’ in Proc. 23rd Annu. Int. Conf. IEEE Eng. Med. Biol. Soc., vol. 4, Oct. 2001, pp. 3721–3723.
[12] Kan Luo, Jianqing Li, Zhigang Wang, Alfred Cuschieri. Patient-Specific Deep Architecture Model for ECG Classification, Journal of Healthcare Engineering. 2017
[13] V. N. Murthy and S. Jayaraman, ‘‘Human electrocardiogram for biometrics using DTW and FLDA,’’ in Proc. 20th Int. Conf. Pattern Recognit. (ICPR), Aug. 2010, pp. 3838–3841
[14] L. Scalise, Y. Velchev, K. Tonchev, and I. Paliy, ‘‘Non contact heart monitoring,’’ in Advances in Electrocardiograms—Methods and Analysis, R. M. Millis, Ed. Rijeka, Croatia: InTech, 2012
[15] D. Obeid, G. Zaharia, S. Sadek, and G. El Zein, ‘‘Microwave Doppler radar for heartbeat detection vs electrocardiogram,’’ Microw. Opt. Technol. Lett., vol. 54, no. 11, pp. 2610–2617, 2012.
[16] S. Bounyong, M. Yoshioka, and J. Ozawa, ‘‘Monitoring of a driver’s heart rate using a microwave sensor and template-matching algorithm,’’ in Proc. IEEE Int. Conf. Consum. Electron. (ICCE), Jan. 2017, pp. 43–44.
[17] F. Lin, C. Song, Y. Zhuang, W. Xu, C. Li, and K. Ren, “Cardiac scan: A non-contact and continuous heart-based user authentication system,” in Proc. ACM ACM Int. Conf. Mobile Comput. Netw., 2017, pp. 315–328.
[18] J. Pan and W. J. Tompkins, “A real-time QRS detection algorithm,” IEEE Trans. Biomed. Eng., vol. BME-32, pp. 230-236, 1985.
[19] Philip de Chazal ; Richard B. Reilly. A Patient-Adapting Heartbeat Classifier Using ECG Morphology and Heartbeat Interval Features. IEEE Transactions on Biomedical Engineering. 20 November 2006.
[20] Philip de Chazal ; M. O'Dwyer ; R.B. Reilly. Automatic classification of heartbeats using ECG morphology and heartbeat interval features. IEEE Transactions on Biomedical Engineering. 21 June 2004.
[21] Meng-Hsi Wu ; Emily J. Chang ; Tzu-Hsuan Chu. Personalizing a Generic ECG Heartbeat Classification for Arrhythmia Detection: A Deep Learning Approach. 2018 IEEE Conference on Multimedia Information Processing and Retrieval (MIPR). 10-12 April 2018
[22] M. A. Tinati and B. Mozaffary, “A wavelet packets approach to electrocardiograph baseline drift cancellation,” International Journal of Biomedical Imaging, vol. 2006, p. 97157, 2006.
[23] Wenbo Zhang ; Limin Yu ; Lishan Ye ; Weifen Zhuang ; Fei Ma. ECG Signal Classification with Deep Learning for Heart Disease Identification. 2018 International Conference on Big Data and Artificial Intelligence (BDAI). 22-24 June 2018
[24] Lin Zhou ; Yan Yan ; Xingbin Qin ; Chan Yuan ; Dashun Que ; Lei Wang. Deep learning-based classification of massive electrocardiography data. 2016 IEEE Advanced Information Management, Communicates, Electronic and Automation Control Conference (IMCEC). 3-5 Oct. 2016
[25] Guo‐Liang Li MD, Ardan M. Saguner MD, Deniz Akdis MD, Guy Hugues Fontaine MD, PhD, HDR, FHRS. Value of a novel 16‐lead High‐Definition ECG machine to detect conduction abnormalities in structural heart disease. ELECTROPHYSIOLOGY. 30 March 2018
[26] Antoni Burguera. Fast QRS Detection and ECG Compression Based on Signal Structural Analysis. IEEE Journal of Biomedical and Health Informatics. 12 January 2018.
[27] V.Mondéjar-Guerra, J.Novo, J.Rouco, M.G.Penedo, M.Ortega. Heartbeat classification fusing temporal and morphological information of ECGs via ensemble of classifiers. Biomedical signal processing and control. 2019/01.
[28] Shu Lih Oh, Eddie Y.K.Ng, Ru San Tan, U. Rajendra Acharya. Automated diagnosis of arrhythmia using combination of CNN and LSTM techniques with variable length heart beats. Computers in Biology and Medicine. 2018/06.
[29] ÖzalYıldırım, Paweł Pławiak, Ru-SanTan, U. RajendraAcharya. Arrhythmia detection using deep convolutional neural network with long duration ECG signals. Computers in Biology and Medicine. 2018/11.
[30] I. Odinaka et al., ‘‘ECG biometrics: A robust short-time frequency analysis,’’ in Proc. IEEE Int. Workshop Inf. Forensics Secur., Dec. 2010, pp. 1–6.
[31] Qingxue Zhang ; Dian Zhou ; Xuan Zeng. HeartID: A Multiresolution Convolutional Neural Network for ECG-Based Biometric Human Identification in Smart Health Applications. IEEE Access. 2017/05
[32] Jonathan Rubin, Saman Parvaneh, Asif Rahman, Bryan Conroy, Saeed Babaeizadeh. Densely connected convolutional networks for detection of atrialfibrillation from short single-lead ECG recordings. Journal of electrocardiology. 2018/12.
[33] Bo Wu, Gongping Yang, Lu Yang, Yilong Yin. Robust ECG Biometrics Using Two-Stage Model. 2018 24th International Conference on Pattern Recognition (ICPR). August 20-24, 2018.
[34] A.S.Alvarado,C.Lakshminarayan,J.C.Principe,Time-basedcompressionandclassificationofheartbeats. IEEE Trans. Biomed. Eng. 59(6), 1641–1648 (2012)
[35] A.E. Zadeh, A. Khazaee, V. Ranaee, Classification of the electrocardiogram signals using supervised classifiers and efficient features. Comput. Methods Prog. Biomed. 99(2), 179–194 (2010)
[36] C.Kamath, ECGbeat classification using featuresextracted from Teager energyfunctionsintime and frequency domains. IET Signal Process. 5(6), 575–581 (2011).
[37] T.Mar,S.Zaunseder, J.P.Martnez, M.Llamedo, R.Poll, Optimization of ECG classification by means of feature selection, IEEETrans. Biomed. Eng.58(8)(2011)2168–2177
[38] A.S. Al-Fahoum, I.Howitt, Combined wavelet transformation and radial basis neural networks for classifying life-threatening cardiac arrhythmias, Med.Biol.Eng.Comput.37 (5)(1999)566–573, http://dx.doi.org/10.1007/BF02513350.
[39] Elmehdi Benmalek ; Jamal Elmhamdi. Arrhythmia ECG signal analysis using non parametric time-frequency technique. 2015 International Conference on Electrical and Information Technologies (ICEIT). 25-27 March 2015.
[40] Stalin Subbiah , Rajkumar Patro , Subbuthai P. "Feature Extraction and Classification for ECG Signal Processing based on Artificial Neural Network and Machine Learning Approach." International Conference on Inter Disciplinary Research in Engineering and Technology :50-57, 2015.
[41] C. Ye, B.V.K.V. Kumar, M.T. Coimbra, Heartbeat classification using morphological and dynamic features of ECG signals. IEEE Trans. Biomed. Eng. 59(10), 2930–2941 (2012)
[42] D. Giri, U.R. Acharya, R.J. Martis, S.V. Sree, T.C. Lim, T. Ahamed, J.S. Suri, Automated diagnosis of coronary artery disease affected patients using LDA, PCA, ICA and discrete wavelet transform. Knowl. Based Syst. 37, 274–282 (2013)
[43] Robin Tan, Marek Perkowski. Toward Improving Electrocardiogram (ECG) Biometric Verification using Mobile Sensors: A Two-Stage Classifier Approach. Sensor. 2017/02
[44] Li, Hongqiang, et al. "A New ECG Signal Classification Based on WPD and ApEn Feature Extraction." Circuits, Systems, and Signal Processing 35.1 : 339-352, 2016.
[45] Li, Hongqiang, et al. "Novel ECG Signal Classification Based on KICA Nonlinear Feature Extraction." Circuits, Systems, and Signal Processing 35.4 : 1187-1197, 2016.
[46] Y. Kutlu, D. Kuntalp, Feature extraction for ECG heartbeats using higher order statistics of WPD coefficients. Comput. Methods Prog. Biomed. 105(3), 257–267 (2012)
[47] Yann LeCun, Leon Bottou, Yoshua BEngio, Patrick Haffner. Gradient-Based Learning Applied to Document Recognition. PROC. OF THE IEEE, 1998/11
[48] Nair, V., & Hinton, G. E. (2010). Rectified linear units improve restricted boltzmann machines. In Proceedings of the 27th international conference on machine learning (ICML-10) (pp. 807-814).
[49] Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2012). Imagenet classification with deep convolutional neural networks. In Advances in neural information processing systems (pp. 1097-1105).
[50] K. Simonyan and A. Zisserman, “Very deep convolutional networks for large-scale image recognition,” in Proc. Int. Conf. Learning Representations (ICLR), 2015.
[51] C. Szegedy, W. Liu, Y. Jia, P. Sermanet, S. Reed, D. Anguelov, D. Erhan, V. Vanhoucke, and A. Rabinovich, “Going deeper with convolutions,” in Proc. IEEE Conf. Computer Vision and Pattern Recognition (CVPR), 2015
[52] He, Kaiming, et al. "Deep residual learning for image recognition." Proceedings of the IEEE conference on computer vision and pattern recognition. 2016.
[53] J. L. Elman, “Finding structure in time,” Cognitive science, vol. 14, no. 2, pp. 179–211, 1990.
[54] S. Hochreiter, J. Schmidhuber, “Long short-term memory”, Neural Comput., 9 (1997) 1735–1780.
[55] Mnih, Volodymyr, Nicolas Heess, and Alex Graves. "Recurrent models of visual attention." Advances in neural information processing systems. 2014.
[56] Bahdanau, Dzmitry, Kyunghyun Cho, and Yoshua Bengio. "Neural machine translation by jointly learning to align and translate." arXiv preprint arXiv:1409.0473 (2014).
[57] Xu, Kelvin, et al. "Show, attend and tell: Neural image caption generation with visual attention." International conference on machine learning. 2015.
[58] Vaswani, Ashish, et al. "Attention is all you need." Advances in neural information processing systems. 2017.
[59] Fei Wang, Mengqing Jiang, Chen Qian, Shuo Yang, Cheng Li, Honggang Zhang, Xiaogang Wang, Xiaoou Tang. Residual Attention Network for Image Classification. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). 21-26 July 2017.
[60] XuK, BaJ, and CourvilleA KirosR. "Show, attendandtell: Neural imagecaptiongenerationwithvisualattention." Proceedings oftheInternationalConferenceonMachineLearning. Lille, France2048 (2015): 2057.
[61] Long Chen, Hanwang Zhang, Jun Xiao, Liqiang Nie, Jian Shao, Wei Liu, Tat-Seng Chua. SCA-CNN: Spatial and Channel-wise Attention in Convolutional Networks for Image Captioning. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) . 21-26 July 2017.
[62] M. Zihlmann, D. Perekrestenko, and M. Tschannen, “Convolutional recurrent neural networks for electrocardiogram classification,” arXiv preprint arXiv:1710.06122, 2017.
[63] Fan, Xiaomao, et al. "Multiscaled fusion of deep convolutional neural networks for screening atrial fibrillation from single lead short ECG recordings." IEEE journal of biomedical and health informatics 22.6 (2018): 1744-1753.
[64] Murugesan, Balamurali, et al. "ECGNet: Deep Network for Arrhythmia Classification." 2018 IEEE International Symposium on Medical Measurements and Applications (MeMeA). IEEE, 2018.
[65] Yildirim, Ozal, et al. "A new approach for arrhythmia classification using deep coded features and LSTM networks." Computer methods and programs in biomedicine 176 (2019): 121-133.
[66] da Silva Luz, Eduardo Jose, et al. "Learning deep off-the-person heart biometrics representations." IEEE Transactions on Information Forensics and Security 13.5 (2017): 1258-1270.
[67] Clifford, Gari D., et al. "AF Classification from a short single lead ECG recording: the PhysioNet/Computing in Cardiology Challenge 2017." 2017 Computing in Cardiology (CinC). IEEE, 2017.
[68] Gari Clifford, Chengyu Liu, Benjamin Moody, Ikaro Silva, Qiao Li, Alistair Johnson, Roger Mark. AF Classification from a Short Single Lead ECG Recording: the PhysioNet Computing in Cardiology Challenge 2017. Computing in Cardiology (Rennes: IEEE). 2017
[69] A.L. Goldberger, et al. PhysioBank, PhysioToolkit, and PhysioNet. Components of a New Research Resource for Complex Physiologic Signals, 101 (23) (2000), pp. e215-e220
[70] Oktay, Ozan, et al. "Attention u-net: Learning where to look for the pancreas." arXiv preprint arXiv:1804.03999 (2018).
[71] Ronneberger, Olaf, Philipp Fischer, and Thomas Brox. "U-net: Convolutional networks for biomedical image segmentation." International Conference on Medical image computing and computer-assisted intervention. Springer, Cham, 2015.
[72] Chen, Liang-Chieh, et al. "Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected crfs." IEEE transactions on pattern analysis and machine intelligence 40.4 (2017): 834-848.
[73] Philip Warrick, Masun Nabhan Homsi, PeriGen. Inc., Montreal, Canada, Simon Bolivar University. Cardiac Arrhythmia Detection from ECG Combining Convolutional and Long Short-Term Memory Networks. arXiv:1801.10033. 2018/01
[74] Fabio Hernández, Dilio Méndez, Lusvin Amado, Miguel Altuve. Atrial Fibrillation Detection in Short Single Lead ECG Recordings Using Wavelet Transform and Artificial Neural Networks. 2018 40th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC). 2018/07
[75] Oh, Shu Lih, et al. "Automated beat-wise arrhythmia diagnosis using modified U-net on extended electrocardiographic recordings with heterogeneous arrhythmia types." Computers in biology and medicine 105 (2019): 92-101.