臺灣博碩士論文加值系統

心電圖跳動辨識（Electrocardiogram beat discrimination）在臨床診斷不同心臟疾病方面扮演著相當重要的角色。即使目前已有釵h關於心電圖跳動辨識的方法被提出來，但若考慮到各種不同的課題，這些方法之間仍然具有相當多可以改善的空間。本研究在發展心電圖跳動辨識方法的過程中，將納入數個重要的議題，其中包括辨識率、抗雜訊能力、以及特徵維度等。
在第二章中，心電圖訊號首先透過離散小波轉換，被分解成不同的次頻（subband）成份，其統計及型態學上的相關特徵則被用來描述原來的心電圖訊號。隨後則利用機率式類神經網路（PNN）進行各種心臟異常搏動的辨識。實驗結果顯示，本方法不但可達到99.65%的辨識率，對於各種不同心搏類別的鑑識率也都達到99%以上。
在第三章中，除了離散小波轉換之外，另外加入了高階統計（Higher order statistics）來描述心電圖訊號，用以提昇心搏辨識過程中抗雜訊的能力（Noise- resistibility）。分類器方面，則採用前饋式倒傳遞類神經網路（Feed-forward back-propagation neural network）。當我們在實驗過程中每筆心電圖紀錄都擷取一種以上的心搏類別作為實驗資料時，資料的結構將變得更為複雜。但實驗結果顯示，即使針對這種更為複雜的心電圖辨識問題，本方法仍可達到97.5%以上的辨識率。
在第四章中，我們針對第三章中所提出的特徵，利用四種非線性特徵選取方式進行特徵維度的降低。實驗結果顯示，同時考量特徵與特徵，以及特徵與類別之間相關性的NCBF方法，在針對分類的議題下，對於選取最具代表性特徵的能力上顯然優於其他方法。當特徵維度降低到只有八個特徵時，本論文所提出的心電圖辨識方法（第三章）仍可維持辨識率達96.34%。
最後，我們考慮針對另一種以訊號分解方法為基礎的心電圖辨識方法進行比較，該方法在訊號分解方面採行的是獨立成份分析法（Independent component analysis, ICA），分類器則是支持向量機（Support vector machine, SVM），我們以ICA-SVM表示該方法。實驗結果顯示，穩態（Stationary）的雜訊包括高斯白雜訊以及電力線干擾對於兩種心電圖辨識方法的辨識結果影響都不大。相較於ICA-SVM方法而言，我們在第三章中所提出的心電圖辨識方法對於非穩態雜訊如基線漂移（Baseline wander）及肌電訊號干擾（Muscle artifact）有更好的容忍能力。當這兩種非穩態雜訊對心電圖訊號的訊雜比（SNR）降至10 dB時，本論文所提出的方法仍可達到90%以上的準確率。

Electrocardiogram (ECG) beat discrimination plays an important role in the clinical diagnosis of heart diseases. Although many ECG beat classification methods have been provided in the literature, there still leave room for improvement in view of different issues. Several significant issues, including recognition rates, noise-resistibility, and feature dimension, are considered in the dissertation for the development of an effective and efficient ECG beat classifier.
In Chapter II, the discrete wavelet transformation is employed to decompose the ECG signals into different subband components. Statistical and morphological features are extracted to characterize the ECG signals. A probabilistic neural network (PNN) proceeds to discriminate different pathological heartbeat types. The results demonstrate that it provides a promising accuracy of 99.65%, with equally well recognition rates of over 99% throughout all heartbeat types in this study.
In Chapter III, higher order statistics is recruited to accompany with the discrete wavelet decomposition to characterize the ECG signals as an attempt to elevate the noise-resistibility of the heartbeat discrimination. A feed-forward back-propagation neural network (FFBNN) is employed as classifier. More than 97.5% discrimination rate is achieved with a more complicated experimental profile in which multiple beat types are selected from each of the records for study.
In Chapter IV, four nonlinear feature selection methods including Relief-F, two nonlinear correlation based filters (NCBFs), and symmetrical uncertainty feature-class only (SUFCO), are utilized to reduce the dimension of features mentioned in Chapter III. The results demonstrate that two NCBFs based on both feature-feature and feature-class correlation measures outperform the other methods. As high as 96.34% accuracies can be retained even with only eight features.
At last, comparison between the proposed methods in Chapter II with another ECG beat discrimination based on independent component analysis and support vector machine (ICA-SVM) method is demonstrated in Chapter V. The results show that both ECG beat classification methods are insensitive to the stationary artifacts including white Gaussian noise and power line interference. The proposed method is especially tolerant to non-stationary artifacts baseline wander and muscle artifacts when compared to ICA-SVM. More than 90% accuracy can be retained with the proposed method even when the SNR is decreases to 10 dB.

中文摘要
ABSTRACT
TABLE OF CONTENTS
LIST OF FIGURES
LIST OF TABLES
ABBREVIATION
I. INTRODUCTION
1.1 Research Motivation
1.2 Issue Description
1.3 Organization
II. ELECTROCARDIOGRAM BEAT CLASSIFICATION BASED ON WAVELET TRANSFORMATION AND PROBABILISTIC NEURAL NETWORK
2.1 Introduction
2.2 Proposed Method
2.2.1 Discrete wavelet transformation
2.2.2 Feature extraction
2.2.3 Normalization of feature vectors
2.2.4 Classification using the probabilistic neural network
2.3 Experiment Design
2.4 Experimental Results
2.5 Discussions
2.5.1 Effects of noise on classification
2.5.2 Comparison of PNN with feed-forward backpropagation neural network
2.5.3 Comparison of the proposed method to other ECG beat classification systems
2.6 Conclusion
III. NOISE-TOLERANT ELECTROCARDIOGRAM BEAT CLASSIFICATION BASED ON HIGHER ORDER STATISTICS OF SUBBAND COMPONENTS
3.1 Introduction
3.2 Databases
3.2.1 MIT-BIH arrhythmia database
3.2.2 Sleep heart health study polysomnography database
3.2.3 MIT-BIH noise stress test database
3.3 Theoretic Background of the Method
3.3.1 Discrete wavelet transform
3.3.2 High order statistics
3.4 Proposed Method
3.4.1 Feature extraction based on HOS of subband signals
3.4.2 Normalization
3.4.3 Feature selection
3.4.4 Non-linear neural classifier
3.5 Contaminating Noises for Experiments
3.5.1 White Gaussian noise
3.5.2 Baseline wander
3.5.3 Power line interference
3.5.4 Muscle artifacts
3.6 Experimental Design
3.6.1 Experimental procedure
3.6.2 Different strategies of ECG sample selection
3.6.3 Feature dimension reduction
3.6.4 Influences of different noise artifacts on ECG beat classification
3.7 Results and Discussion
3.7.1 Effects of different sample selection profiles
3.7.2 Reliability of the proposed feature selection method
3.7.3 Influence of noise artifacts on classification
3.7.4 Comparison of the proposed method to other ECG beat classification systems
3.8 Conclusion
IV. SELECTION OF EFFECTIVE ECG FEATURES BASED ON NONLINEAR CORRELATIONS
4.1 Introduction
4.2 Feature Selection Methods Employed in this Study
4.2.1 Nonlinear correlation-based filter
4.2.2 Relief-F
4.3 Experimental Design
4.3.1 Database and preparation for signal analysis
4.3.2 Feature extraction based on HOS of subband signals
4.3.3 Normalization
4.3.4 Feature selection
4.3.5 Nonlinear classifier
4.4 Results and Discussion
4.5 Conclusion
V. COMPARISON OF TWO ECG BEAT CLASSIFIERS BASED ON DIFFERENT SIGNAL DECOMPOSITION METHODS
5.1 Introduction
5.2 Methods
5.2.1 HOS-DWT-FFBNN method
5.2.2 ICA-SVM method
5.3 Experimental Design
5.3.1 Material
5.3.2 Feature dimension reduction
5.3.3 Noise Resistibility
5.4 Results and Discussion
5.4.1 Capability of feature dimension reduction
5.4.2 Capability of noise resistibility
5.5 Conclusion
VI. CONCLUSION AND PERSPECTIVES
6.1 Conclusion
6.2 Future Researches
REFERENCES
VITA

[1]S. Osowski, L. T. Hoai, and T. Markiewicz, "Support vector machine-based expert system for reliable heartbeat recognition," IEEE Transactions on Biomedical Engineering, vol. 51, pp. 582-589, 2004.
[2]S. Osowski and T. H. Linh, "ECG beat recognition using fuzzy hybrid neural network," IEEE Transactions on Biomedical Engineering, vol. 48, pp. 1265-1271, 2001.
[3]I. Guler and E. D. Ubeyli, "A modified mixture of experts network structure for ECG beats classification with diverse features," Engineering Applications of Artificial Intelligence, vol. 18, pp. 845-856, 2005.
[4]M. Engin, "ECG beat classification using neuro-fuzzy network," Pattern Recognition Letters, vol. 25, pp. 1715-1722, 2004.
[5]M. Lagerholm, C. Peterson, G. Braccini, L. Edenbrandt, and L. Sornmo, "Clustering ECG complexes using Hermite functions and self-organizing maps," IEEE Transactions on Biomedical Engineering, vol. 47, pp. 838-848, 2000.
[6]I. Jekova, G. Bortolan, and I. Christov, "Assessment and comparison of different methods for heartbeat classification," Medical Engineering & Physics, vol. 30, pp. 248-257, 2008.
[7]S. W. Chen, "Complexity-measure-based sequential hypothesis testing for real-time detection of lethal cardiac arrhythmias," Eurasip Journal on Advances in Signal Processing, pp. 8, 2007.
[8]H. G. Hosseini, D. Luo, and K. J. Reynolds, "The comparison of different feed forward neural network architectures for ECG signal diagnosis," Medical Engineering & Physics, vol. 28, pp. 372-378, 2006.
[9]I. Christov, G. Gomez-Herrero, V. Krasteva, I. Jekova, A. Gotchev, and K. Egiazarian, "Comparative study of morphological and time-frequency ECG descriptors for heartbeat classification," Medical Engineering & Physics, vol. 28, pp. 876-887, 2006.
[10]M. G. Tsipouras, D. I. Fotiadis, and D. Sideris, "An arrhythmia classification system based on the RR-interval signal," Artificial Intelligence in Medicine, vol. 33, pp. 237-250, 2005.
[11]P. de Chazal and R. B. Reilly, "Automatic classification of ECG beats using waveform shape and heart beat interval features," presented at IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP ''03), 2003.
[12]Y. H. Hu, S. Palreddy, and W. J. Tompkins, "A patient-adaptable ECG beat classifier using a mixture of experts approach," IEEE Transactions on Biomedical Engineering, vol. 44, pp. 891-900, 1997.
[13]R. E. Challis and R. I. Kitney, "Biomedical signal processing (in four parts). Part 1. Time-domain methods," Medical and Biological Engineering and Computing, vol. 28, pp. 509-24, 1990.
[14]J. Moraes, M. O. Seixas, F. N. Vilani, and E. V. Costa, "A real time QRS complex classification method using Mahalanobis distance," Computers in Cardiology, 2002, pp. 201-204, 2002.
[15]L. Khadra, A. S. Al-Fahoum, and S. Binajjaj, "A quantitative analysis approach for cardiac arrhythmia classification using higher order spectral techniques," IEEE Transactions on Biomedical Engineering, vol. 52, pp. 1840-1845, 2005.
[16]K. Minami, H. Nakajima, and T. Toyoshima, "Real-time discrimination of ventricular tachyarrhythmia with Fourier-transform neural network," IEEE Transactions on Biomedical Engineering, vol. 46, pp. 179-185, 1999.
[17]S. N. Yu and Y. H. Chen, "Electrocardiogram beat classification based on wavelet transformation and probabilistic neural network," Pattern Recognition Letters, vol. 28, pp. 1142-1150, 2007.
[18]E. D. Ubeyli, "ECG beats classification using multiclass support vector machines with error correcting output codes," Digital Signal Processing, vol. 17, pp. 675-684, 2007.
[19]G. Krishna Prasad and J. S. Sahambi, "Classification of EGG arrhythmias using multi-resolution analysis and neural networks," presented at IEEE TENCON 2003: Confernce on Covergent Technologies for the Asia-Pacific Region, Bangalore, India, 2003.
[20]P. K. Dash, "Electrocardiogram monitoring," Indian J. Anaesth, vol. 46, pp. 251-260, 2002.
[21]J. Jakubowski, K. Kwiatos, A. Chwaleba, and S. Osowski, "Higher order statistics and neural network for tremor recognition," IEEE Transactions on Biomedical Engineering, vol. 49, pp. 152-159, 2002.
[22]S. N. Yu and K. T. Chou, "Selection of significant independent components for ECG beat classification," Expert Systems with Applications, 2007.
[23]I. Guler and E. D. Ubeyli, "ECG beat classifier designed by combined neural network model," Pattern Recognition, vol. 38, pp. 199-208, 2005.
[24]P. S. Addison, J. N. Watson, G. R. Clegg, M. Holzer, F. Sterz, and C. E. Robertson, "Evaluating arrhythmias in ECG signals using wavelet transforms," IEEE Engineering in Medicine and Biology Magazine, vol. 19, pp. 104-109, 2000.
[25]A. S. Al-Fahoum and I. Howitt, "Combined wavelet transformation and radial basis neural networks for classifying life-threatening cardiac arrhythmias," Medical & Biological Engineering & Computing, vol. 37, pp. 566-573, 1999.
[26]P. D. Wasserman, Advanced Methods in Neural Computing: John Wiley & Sons, Inc. New York, NY, USA, 1993.
[27]J. F. Scholl, J. R. Agre, L. P. Clare, and M. C. Gill, "Low-power impulse signal classifier using the Haar wavelet transform," presented at Proc. SPIE, 1999.
[28]S. K. Goumas, M. E. Zervakis, and G. S. Stavrakakis, "Classification of washing machines vibration signals using discretewavelet analysis for feature extraction," IEEE Transactions on Instrumentation and Measurement, vol. 51, pp. 497-508, 2002.
[29]L. Rutkowski, "Adaptive probabilistic neural networks for pattern classification in time-varying environment," IEEE Transactions on Neural Networks, vol. 15, pp. 811-827, 2004.
[30]A. L. Goldberger, L. A. N. Amaral, L. Glass, J. M. Hausdorff, P. C. Ivanov, R. G. Mark, J. E. Mietus, G. B. Moody, C. K. Peng, and H. E. Stanley, "PhysioBank, PhysioToolkit, and PhysioNet - Components of a new research resource for complex physiologic signals," Circulation, vol. 101, pp. E215-E220, 2000.
[31]Physiobank Archieve Index, MIT-BIH Arrhythmia Database: http://www.physionet.org/physiobank/database/mitdb/ (Accessed: 29 June, 2008)
[32]Z. Jie, "Automatic detection of premature ventricular contraction using quantum neural networks," presented at Third IEEE Symposium on Bioinformatics and Bioengineering Proceedings, 2003.
[33]F. M. Charbonnier, "External defibrillators and emergency external pacemakers," Proceedings of the IEEE, vol. 84, pp. 487-499, 1996.
[34]S. Haykin, Neural Networks: A Comprehensive Foundation. Englewood Cliffs, NJ: Prentice-Hall, 1994.
[35]G. E. Oien, N. A. Bertelsen, T. Eftestol, and J. H. Husoy, "ECG rhythm classification using artificial neural networks," presented at IEEE Digital Signal Processing Workshop Proceedings, 1996.
[36]W. Jiang and S. G. Kong, "Block-based neural networks for personalized ECG signal classification," IEEE Transactions on Neural Networks, vol. 18, pp. 1750-1761, 2007.
[37]K. Polat and S. Gunes, "Detection of ECG Arrhythmia using a differential expert system approach based on principal component analysis and least square support vector machine," Applied Mathematics and Computation, vol. 186, pp. 898-906, 2007.
[38]J. P. Martinez, R. Almeida, S. Olmos, A. P. Rocha, and P. Laguna, "A wavelet-based ECG delineator: Evaluation on standard databases," IEEE Transactions on Biomedical Engineering, vol. 51, pp. 570-581, 2004.
[39]J. S. Sahambi, S. N. Tandon, and R. K. P. Bhatt, "Using wavelet transforms for ECG characterization - An on-line digital signal processing system," IEEE Engineering in Medicine and Biology Magazine, vol. 16, pp. 77-83, 1997.
[40]F. Fleuret, "Fast binary feature selection with conditional mutual information," Journal of Machine Learning Research, vol. 5, pp. 1531-1555, 2004.
[41]L. Yu and H. Liu, "Feature Selection for High-Dimensional Data: A Fast Correlation-Based Filter Solution," presented at Proceedings of The Twentieth International Conference on Machine Leaning (ICML-2003), Washington, D.C., USA, 2003.
[42]R. Battiti, "Using Mutual Information for Selecting Features in Supervised Neural-Net Learning," IEEE Transactions on Neural Networks, vol. 5, pp. 537-550, 1994.
[43]Physiobank Archieve Index, Sleep Heart Health Study Polysomnography Database: http://www.physionet.org/physiobank/database/shhpsgdb/ (Accessed: 29 June, 2008)
[44]Physiobank Archieve Index, MIT-BIH Noise Stress Test Database: http://www.physionet.org/physiobank/database/nstdb/ (Accessed: 29 June, 2008)
[45]R. O. Duda, P. E. Hart, and D. G. Stork, "Pattern Classification," Second ed. New York: John Wiley & Sons, Inc., 2001, pp. 117-124.
[46]F. M. Ham and S. Han, "Classification of cardiac arrhythmias using fuzzy ARTMAP," IEEE Transactions on Biomedical Engineering, vol. 43, pp. 425-429, 1996.
[47]K. Daqrouq, "ECG Baseline Wandering Reduction Using Discrete Wavelet Transform," Asian Journal of Information Technology, vol. 4, pp. 989-995, 2005.
[48]B. Boucheham, Y. Ferdi, and M. C. Batouche, "Piecewise linear correction of ECG baseline wander: a curve simplification approach," Computer Methods and Programs in Biomedicine, vol. 78, pp. 1-10, 2005.
[49]J. A. Van Alste and T. S. Schilder, "Removal of Base-Line Wander and Power-Line Interference from the ECG by an Efficient FIR Filter with a Reduced Number of Taps," IEEE Transactions on Biomedical Engineering, vol. 32, pp. 1052-1060, 1985.
[50]S. M. M. Martens, M. Mischi, S. G. Oei, and J. W. M. Bergmans, "An improved adaptive power line interference canceller for electrocardiography," IEEE Transactions on Biomedical Engineering, vol. 53, pp. 2220-2231, 2006.
[51]P. P. Vaidyanathan, Multirate systems and filter banks: Prentice-Hall, Inc. Upper Saddle River, NJ, USA, 1993.
[52]S. F. Quan, B. V. Howard, C. Iber, J. P. Kiley, F. J. Nieto, G. T. O''Connor, D. M. Rapoport, S. Redline, J. Robbins, J. M. Samet, and P. W. Wahl, "The sleep heart health study: Design, rationale, and methods," Sleep, vol. 20, pp. 1077-1085, 1997.
[53]H. Liu, "Evolving feature selection," IEEE Intelligent Systems, vol. 20, pp. 64-76, 2005.
[54]M. Robnik-Sikonja and I. Kononenko, "Theoretical and empirical analysis of ReliefF and RReliefF," Machine Learning, vol. 53, pp. 23-69, 2003.
[55]Y. Li, Z.-F. Wu, J.-M. Liu, and Y.-Y. Tang, "Efficient feature selection for high-dimensional data using two-level filter," presented at Proceeding of 2004 International Conference on Machine Learning and Cybernetics (ICMLC), 2004.
[56]Y.-H. Chen and S.-N. Yu, "Subband features based on higher order statistics for ECG beat classification," presented at 29th Annual International Conference of IEEE-EMBS, Engineering in Medicine and Biology Society, EMBC''07, Lyon, France, 2007.
[57]R. O. Duda, P. E. Hart, and D. G. Stork, "Parzen Windows," in Pattern Classification, Second ed. New York: John Wiley & Sons, Inc., 2001, pp. 164-174.
[58]S.-N. Yu and Y.-H. Chen, "Selection of Higher Order Subband Features for ECG Beat Classification," presented at 16th European Signal Processing Conference, EUSIPCO 2008, Lausanne, Switzerland, 2008.
[59]A. Hyvarinen, J. Karhunen, and E. Oja, "Independent Component Analysis," IEEE Transactions on Biomedical Engineering, vol. 14, pp. 21-30, 2000.
[60]M. Owis, A. Youssef, and Y. Kadah, "Characterisation of electrocardiogram signals based on blind source separation," Medical and Biological Engineering and Computing, vol. 40, pp. 557-564, 2002.
[61]A. Hyvarinen, "Fast and robust fixed-point algorithms for independent componentanalysis," IEEE Transactions on Neural Networks, vol. 10, pp. 626-634, 1999.
[62]C. W. Hsu and C. J. Lin, "A comparison of methods for multiclass support vector machines," IEEE Transactions on Neural Networks, vol. 13, pp. 415-425, 2002.
[63]J. Friedman, "Another approach to polychotomous classification," Dept. Statistics, Stanford Univ., Tech. Rep, 1996.