近年來，隨著心血管疾病數量的顯著增加，心電信號的分類研究一直在臨床
診斷中起著非常重要的作用。本文主要分為兩項子研究，第一項子研究主要基於
一維卷積神經網絡(CNN)提出了心電訊號分類的方法。所提出的檢測系統可以組
成分為三個部分:數據前處理、模型建立和分類。通過神經網絡結構對模型進行
訓練，得到分類結果採用6層網絡結構。主要的分類有兩類，分別是心衰竭病患
與控制組，訓練準確度達到97.57%且測試的準確度達到95.31%，明顯優於多個測
試集典型的心電圖分類方法。
第二項子研究除了監測心電圖，它還有評估呼吸活動。呼吸信號本身是不記
錄的，它可以從心電圖中提取出來(EDR)。本文提出了一種估計和消除基線漂移
的算法和獲得的EDR信號估計一個病人的呼吸頻率。我們利用深度學習神經網絡，
通過輸入心跳間隔信號與提取出的呼吸間隔資料來訓練時域和頻域的特徵。我們
使用了卷積神經網絡(CNN)建立模型。我們使用500點的心跳間隔和119點的呼吸
間隔作為輸入的數據。除了討論了兩種方法與其他方法的比較，本文還對不同數
據長度的性能進行了比較。目前的模型在訓練集上正確率高達94.78%，在測試準
確率為89.63%，對於辨別心衰竭病患有良好的效果。

Recently, with the significant increase in the number of cardiovascular diseases, automatic
classification study of ECG signals has always played a very important part in clinical diagnosis
of cardiovascular diseases.In this paper, a 1D convolution neural network (CNN) based
method is proposed to classify ECG signals. The proposed detection system could be composed
of three portions: data pre-processing, model-establishment and classification. Afterwards,
through the neural network structure to train the model and get the classification result
by using 6-layer network architecture. The recognition accuracy between CHF and Control is
up to 97.57% for training set, and 95.31% for testing set,significantly outperforming several
typical ECG classification methods.
Besides monitoring the ECG, it is helpful to assess respiratory activity. If the respiration
signal itself is not recorded, it can be extracted from the ECG (i.e. ECG derived respiration,
EDR). In this paper we present a algorithm for estimation and removal of baseline wander
noise and obtaining the EDR signal for estimation of a patients respiratory rate. The estimated
baseline is interpolated from the ECG signal at midpoints between each detected R-wave. We
employed deep learning neural network to train the features by input RR interval signal and
EDR signals in both time and frequency domain. We used convolution neural network (CNN)
to build the model.We used RR interval of 500 points and breathing interval of 119 points as
input data. After two layers of convolution layer and two layers of pooling layer, we passed
through the structure of concatenate layer, flatten layer, dense layer and drop out layer to the
final output of a total of 9 layers. The comparisons of two approaches were discussed in
this thesis, also we compared different data length performance. With our current model, the
recognition accuracy between CHF and Control is up to 94.78% for training set, and 89.63%
for testing set.

1 Introduction 1
1.1 Backgrounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Main Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.4 Thesis Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2 Pathological Knowledge and Methodologies 7
2.1 Overview of ECG Signal . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1.1 Electrophysiology . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1.2 ECG Morphology . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2 Congestive Heart Failure (CHF) . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2.1 Cardiac Arrhythmia . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2.2 Congestive Heart Failure . . . . . . . . . . . . . . . . . . . . . . . . 15
2.3 ECG-Derived Respiration (EDR) . . . . . . . . . . . . . . . . . . . . . . . . 17
2.3.1 Physiology of Respiration . . . . . . . . . . . . . . . . . . . . . . . 17
2.3.2 ECG-Derived Respiration (EDR) . . . . . . . . . . . . . . . . . . . 17
2.4 Convolution Neural Network (CNN) . . . . . . . . . . . . . . . . . . . . . . 19
2.5 Literature Survey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.5.1 Overview of Biomedical Signal Analysis . . . . . . . . . . . . . . . 21
2.5.2 Classification Methods . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.5.3 Neural Network Algorithm for CHF . . . . . . . . . . . . . . . . . . 22
2.5.4 Comparisons of Studies . . . . . . . . . . . . . . . . . . . . . . . . 24
3 Proposed Methodology of Recognizing CHF 27
3.1 Model Establishment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.2 Data Preprocessing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.2.1 Prerequisite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.2.2 RR Interval Extraction . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.2.3 Information of the selected data. . . . . . . . . . . . . . . . . . . . . 36
3.3 Curve of Multiscale Entropy . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.3.1 Gender . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.3.2 Age . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.3.3 Atrial Fibrillation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.3.4 PAC & PVC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.4 Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.4.1 Cross Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.5 A CHF Detection System with 1D CNN . . . . . . . . . . . . . . . . . . . . 42
3.5.1 Database information . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.5.2 Data Preprocessing . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.5.3 1D CNN architecture in CHF detection . . . . . . . . . . . . . . . . 45
3.6 ECG-derived respiration (EDR) signal via Multi-input Deep Learning Network 46
3.6.1 Multi-input Configuration . . . . . . . . . . . . . . . . . . . . . . . 48
3.6.2 Classification with MINN Model . . . . . . . . . . . . . . . . . . . 48
4 Implementation Results and Discussion 53
4.1 Classification with 1D-CNN Model . . . . . . . . . . . . . . . . . . . . . . 55
4.1.1 1D-CNN Results for NTUH Database . . . . . . . . . . . . . . . . . 55
4.1.2 Classification Results for NTUH Database . . . . . . . . . . . . . . . 56
4.2 Classification with EDR Model . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.2.1 Feature Extraction for MINN Model . . . . . . . . . . . . . . . . . . 57
4.2.2 Overall System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.2.3 Other Database Experiments . . . . . . . . . . . . . . . . . . . . . . 62
4.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.3.1 CHF Detection Comparison with Literature . . . . . . . . . . . . . . 64
5 Conclusion and Future Works 67
5.1 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
5.2 Future Works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

[1] “Congestive heart failure,” Sep. 2016. [Online]. Available: http://www.md-health.com/
Congestive-Heart-Failure.html.
[2] “Rise Above Heart Failure,” 2016. [Online]. Available: http://www.heart.org/
HEARTORG/Conditions/HeartFailure/Heart-Failure UCM 002019 SubHomePage.jsp.
[3] “Congestive heart failure (CHF) diagnosis,” Sept. 2015. [Online]. Available:
http://www.healthcommunities.com/congestive-heart-failure/diagnosis.html.
[4] M.-Y. Lee and S.-N. Yu, “Selection of heart rate variability features for congestive
heart failure recognition using support vector machine-based criteria,” in 5th European
Conference of the International Federation for Medical and Biological Engineering.
Springer, 2011, pp. 400–403.
[5] L. Pecchia, P. Melillo, M. Sansone, and M. Bracale, “Discrimination power of short-term
heart rate variability measures for CHF assessment,” IEEE Transactions on Information
Technology in Biomedicine, vol. 15, no. 1, pp. 40–46, 2011.
[6] W. Chen, G. Liu, S. Su, Q. Jiang, and H. Nguyen, “A CHF detection method based on
deep learning with RR intervals,” in 2017 39th Annual International Conference of the
IEEE Engineering in Medicine and Biology Society (EMBC), 2017, pp. 3369–3372.
[7] T. I. S. Kiranyaz and M. Gabbouj, “Real-time patient-specific classification by 1-d convolutional
neural networks,” IEEE Transactions on Biomedical Engineering, vol. 63,
no. 3, pp. 664–675, 2016.
[8] J. E. Hall, Guyton and Hall textbook of medical physiology. Elsevier Health Sciences,
2015.
[9] E. Burns, “Ecg library: Comprehensive, free educational resource covering over 100
topics relevant to emergency medicine and critical care,” 2010. [Online]. Available:
http://lifeinthefastlane.com/ecg-library/.
[10] “Heart,” Jun. 2012. [Online]. Available: http://completesoccertraining.blogspot.tw/
2012/06/heart.html.
[11] “Electrocardiography,” Aug. 2016. [Online]. Available: https://en.wikipedia.org/wiki/
Electrocardiography.
[12] J. W. Hurst, “Naming of the waves in the ECG, with a brief account of their genesis,”
Circulation, vol. 98, no. 18, pp. 1937–1942, 1998.
[13] M. A. Gatzoulis, S. Balaji, S. A. Webber, S. C. Siu, J. S. Hokanson, C. Poile, M. Rosenthal,
M. Nakazawa, J. H. Moller, P. C. Gillette et al., “Risk factors for arrhythmia and
sudden cardiac death late after repair of tetralogy of fallot: a multicentre study,” The
Lancet, vol. 356, no. 9234, pp. 975–981, 2000.
[14] D. Conen, M. Adam, F. Roche, J.-C. Barthelemy, D. F. Dietrich, M. Imboden,
N. K¨unzli, A. von Eckardstein, S. Regenass, T. Hornemann et al., “Premature atrial
contractions in the general population: frequency and risk factors,” Circulation, pp.
CIRCULATIONAHA–112, 2012.
[15] P. A. Van DerWouw, A. C. Brauns, S. E. Bailey, J. E. Powers, and A. A.Wilde, “Premature
ventricular contractions during triggered imaging with ultrasound contrast,” Journal
of the American Society of Echocardiography, vol. 13, no. 4, pp. 288–294, 2000.
[16] P. A. Wolf, R. D. Abbott, and W. B. Kannel, “Atrial fibrillation as an independent risk
factor for stroke: the framingham study.” Stroke, vol. 22, no. 8, pp. 983–988, 1991.
[17] A. Guyton and J. Hall, Textbook of Medical Physiology. Philadelphia: W. B.Saunders
and Company, 1996.
[18] “Forget breath control,” JANUARY 2018. [Online]. Available: https://www.gmangelo.
com/forget-breath-control/
[19] M. Schmidt, J. Passand N Krug, A. Schumann, K.-J. Br, and G. Rose, “Estimation of a
respiratory signal from a single-lead ECG using the 4th order central moments,” Current
Directions in Biomedical Engineering, vol. 1, 09 2015.
[20] G. E. L. P. B. R. Lazaro JL, Alcaine A, “Electrocardiogram derived respiration from qrs
slopes,” in IEEE Engineering in Medicine and Biology Society, 2013, pp. 396–398.
[21] A. Z. George B. Moody, Roger G. Mark and S. Mantero, “Derivation of Respiratory
Signals from Multi-lead ECGs,” Computers in Cardiology, vol. 12, pp. 113–116, 1985.
[22] J. S. D. D. H. R. E. H. W. H. Y. LeCun, B. Boser and L. D. Jackel, “Backpropagation
applied to handwritten zip code recognition,” Neural Comput., vol. 1, no. 4, pp. 541–551,
1989.
[23] C. S. S. B. S. Chandra and S. Jana, “Robust heartbeat detection from multimodal data
via CNN-based generalizable information fusion,” IEEE Trans. Biomed. Eng., vol. 66,
no. 3, pp. 710–717, 2019.
[24] Z. L. Y. Xiang and J. Meng, “Automatic QRS complex detection using two-level convolutional
neural network,” IEEE Trans. Biomed. Eng., vol. 17, no. 1, p. 13, 2018.
[25] V. S. Chouhan and S. S. Mehta, “Threshold-based detection of P and T-wave in ECG
using new feature signal,” International Journal of Computer Science and Network Security,
vol. 8, no. 2, pp. 155–153, 2008.
[26] A. B. S. A. Mahmoud and S. A. Al-Tunaiji, “Six order cascaded power line notch filter
for ECG detection systems with noise shaping,” Circuits, Syst., Signal Process, vol. 33,
no. 88, p. 23852400, 2014.
[27] E. Pasolli and F. Melgani, “Active learning methods for electrocardiographic signal classification,”
IEEE Trans. Inf. Technol. Biomed, vol. 14, no. 6, pp. 1405–1416, 2010.
[28] M. A. Z. Zidelmal, A. Amirou and A. Belouchrani, “QRS detection based on wavelet
coefficients,” Comput. Methods Programs Biomed, vol. 107, pp. 490–496, 2012.
[29] M. A. Kabir and C. Shahnaz, “Denoising of ECG signals based on noise reduction algorithms
in EMD and wavelet domains,” Biomed. Signal Process. Control, vol. 7, no. 5,
pp. 481–489, 2012.
[30] T. Li and Z. Min, “ECG classification using wavelet packet entropy and random forests,”
Entropy, vol. 18, no. 8, p. 285, 2016.
[31] R. C. Y. Ozbay and B. Karlik, “Integration of type-2 fuzzy clustering and wavelet transform
in a neural network based ECG classifier,” Expert Syst. Appl., vol. 38, pp. 1004–
1010, 2011.
[32] X. M. D. C. H. Li, D. Yuan and L. Cao, “Genetic algorithm for the optimization of
features and neural networks in ECG signals classification,” Sci. Rep., vol. 7, no. 41011,
2017.
[33] A. R. H. T. T. S. F. A. Elhaj, N. Salim and T. Ahmed, “Arrhythmia recognition and
classification using combined linear and nonlinear features of ECG signals,” Comput.
Methods Programs Biomed., vol. 127, pp. 52–63, 2016.
[34] C. W. C. Y. C. Yeh and H. J. Lin, “Analyzing ECG for cardiac arrhythmia using cluster
analysis,” Expert Syst. Appl., vol. 39, pp. 1000–1010, 2012.
[35] E. Alickovic and A. Subasi, “Effect of multiscale PCA de-noising in ECG beat classification
for diagnosis of cardiovascular diseases,” Circuits, Syst., vol. 34, no. 3, pp.
513–533, 2015.
[36] O. Yildirim, “A novel wavelet sequence based on deep bidirectional LSTM network
model for ECG signal classification,” Comput. Biol. Med., vol. 96, pp. 189–202, 2018.
[37] C. M. L. K. M. M. A. K. R. R. J. Martis, U. R. Acharya and C. Chakraborty, “Application
of higher order cumulant features for cardiac health diagnosis using ECG signals,” Int.
J. Neural Syst., vol. 23, no. 4, 2013.
[38] B. Y. X.-C. Cao, B.-Q. Chen and W.-P. He, “Combining translationinvariant wavelet
frames and convolutional neural network for intelligent tool wear state identification,”
Comput. Ind., vol. 106, pp. 71–84, 2019.
[39] R. Salloum and C.-C. J. Kuo, “ECG-based biometrics using recurrent neural networks,”
Proc. IEEE Int. Conf. Acoust., Speech Signal Process. (ICASSP), pp. 2062–2066, 2017.
[40] X. S. A. Mostayed, J. Luo and W. Wee, “Classification of 12-lead ECG signals with Bidirectional
LSTM network,” 2018. [Online]. Available: https://arxiv.org/abs/1811.02090
[41] J. Z. P. G. J. L. C. Zhang, G. Wang and H. Yang, “Patient-specific ECG classification
based on recurrent neural networks and clustering technique,” in Proc. 13th IASTED Int.
Conf. Biomed. Eng. (BioMed), 2017, pp. 63–67.
[42] T. I. S. Kiranyaz and M. Gabbouj, “Real-time patient-specific ecg classification by 1-d
convolutional neural networks,” IEEE Trans. Biomed. Eng., vol. 63, no. 3, pp. 664–675,
2016.
[43] Q. Z. D. Li, J. Zhang and X. Wei, “Classification of ECG signals based on 1D convolution
neural network,” in Proc. IEEE 19th Int. Conf E-Health Netw. IEEE, 2017, pp.
1–6.
[44] L. Z. W. Yin, X. Yang and E. Oki, “ECG monitoring system integrated with IR-UWB
radar based on CNN,” IEEE Access, vol. 4, pp. 6344–6351, 2016.
[45] D. K. D. K. D. K. T. J. Jun, H. M. Nguyen and Y.-H. Kim, “ECG arrhythmia
classification using a 2-D convolutional neural network,” 2018. [Online]. Available:
https://arxiv.org/abs/1804.06812
[46] S. T. M. Salem and J.-S. Yuan, “ECG arrhythmia classification using transfer learning
from 2- dimensional deep CNN features,” in Proc. IEEE Biomed. Circuits Syst. Conf.
IEEE, 2018, pp. 1–4.
[47] M. H. C. B. P. Rajpurkar, A. Y. Hannun and A. Y. Ng, “Cardiologist-level
arrhythmia detection with convolutional neural networks,” 2017. [Online]. Available:
https://arxiv.org/abs/1707.01836
[48] M. arlija, F. Jurii, and S. Popovic, “A convolutional neural network based approach to
qrs detection,” 09 2017, pp. 121–125.
[49] F. Chollet et al., “Keras,” https://github.com/fchollet/keras, 2015.
[50] M. Mark H. Ebell M.D., Evidence-Based Diagnosis A Handbook of Clinical Prediction
Rules. Springer New York.
[51] A. L. Goldberger et al., “PhysioBank, PhysioToolkit, and PhysioNet,” Circulation, vol.
101, no. 23, pp. E215–20, 2000.