臺灣博碩士論文加值系統

　　心電圖(ECG)是一種常用的心臟檢測方法，通過捕捉並放大心臟收縮時產生的微量電訊號來觀察心臟是否有疾病。然而，由於心電訊號非常微小，容易受到各種雜訊的干擾，因此在診斷前需通過演算法來消除這些雜訊。本論文提出了一種名為FMHA-AE的架構，用於去除雜訊並重建乾淨的心電圖訊號。FMHA-AE架構的主要目的是提升心電圖訊號的訊雜比(SNRimp)和降低重建後訊號的重構失真度(PRD)。本研究參考了前人的工作，並提出了一種自動編碼器模型，有效去除雜訊並保持心電圖訊號的完整性。
　　實驗結果顯示，FMHA-AE表現優異，相較於傳統方法能顯著提升訊雜比(SNRimp)並降低重構失真度(PRD)。具體來說，使用FMHA-AE架構處理後的心電圖訊號平均SNRimp提升至23.87dB，而平均PRD降低至12.05%。這表明FMHA-AE不僅能有效去除雜訊，還能準確重建心電圖訊號，為臨床心臟病診斷提供了可靠的數據支持。本論文的貢獻在於提出了一種高效的心電圖訊號降噪和重建方法，並通過實驗驗證了其在實際應用中的有效性。

　　Electrocardiogram (ECG) is a commonly used method for heart examination, which observes whether there are heart diseases by capturing and amplifying the tiny electrical signals generated during heart contractions. However, due to the very small size of these electrical signals, they are easily affected by various types of noise.
Therefore, algorithms are needed to eliminate these noises before diagnosis. This thesis proposes a framework called FMHA-AE for denoising and reconstructing clean ECG signals. The main purpose of the FMHA-AE framework is to improve the signal-to-noise ratio(SNRimp) of the ECG signals and reduce the reconstruction distortion(PRD) of the reconstructed signals. This study references previous work and proposes an autoencoder model that effectively removes noise while maintaining the integrity of the ECG signals.
　　Experimental results show that FMHA-AE performs excellently, significantly improving the signal-to-noise ratio(SNRimp) and reducing the reconstruction distortion(PRD) compared to traditional methods. Specifically, the average SNRimp of the ECG signals processed by the FMHA-AE framework is increased to 23.87 dB, while the average PRD is reduced to 12.05%. This indicates that FMHA-AE not only effectively removes noise but also accurately reconstructs ECG signals, providing reliable data support for clinical heart disease diagnosis. The contribution of this thesis lies in proposing an efficient method for ECG signal denoising and reconstruction, and validating its effectiveness in practical applications through experiments.

摘要 i
ABSTRACT ii
目錄 iii
表目錄 v
圖目錄 vi
第一章 緒論 1
1.1 研究動機 1
1.2 心電圖雜訊 2
1.3 研究貢獻 3
1.4 論文架構 4
第二章 文獻回顧與探討 5
2.1 傳統數位訊號處理演算法 5
2.2 DAE架構 6
2.3 Fully-Connection DAE架構 7
2.4 Fully-Convolution DAE架構 8
2.5 CNN-LSTM DAE架構 9
2.6 LMSC DAE架構 10
2.7 CPDAE架構 12
2.8 TCDAE架構 12
2.9 架構比較與探討 15
第三章 FMHA-AE架構設計 17
3.1 神經網路架構 17
3.2 FMHSA Encoder架構設計 19
3.3 MHSA Shortcut架構設計 23
3.4 FMHCA Decoder架構設計 23
第四章 結果與比較 26
4.1 資料集 26
4.2 訓練環境與訓練超參數設置 28
4.3 評估指標 29
4.4 比較論文之實現 29
4.5 實驗數據與結果圖 33
第五章 結論與未來工作 36
參考文獻 37
附錄 40
附錄一、口試委員Q&A 40

[1]“衛生福利部 111 年國人死因統計結果”, [Online].Available: https://www.mohw.gov.tw/cp-16-79055-1.html
[2]“WHO reveals leading causes of death and disability worldwide: 2000-2019”, [Online].Available: https://www.who.int/news-room/fact-sheets/detail/the-top-10-causes-of-death
[3]“心電圖(EKG)檢查與注意事項” https://www.taic.mohw.gov.tw/?aid=509&pid=0&page_name=detail&type=0&iid=2340
[4] “缺血性心臟病的診斷與治療”https://www.vhwc.gov.tw/PageView/RowViewDetail?WebRowsID=26bdc81a-aa43-4157-84d4-e93e54589e86&UnitID=9560bc3c-9b11-43a8-9cd7-89c676199787&CompanyID=e8e0488e-54a0-44bf-b10c-d029c423f6e7
[5]林佳禾。「用於心電訊號之壓縮快捷式降噪自編碼器晶片設計」。碩士論文，國立雲林科技大學電子工程系，2023。https://hdl.handle.net/11296/5e936j。
[6]M. Z. U. Rahman, R. A. Shaik and D. V. R. K. Reddy, “Efficient and simplified adaptive noise cancelers for ECG sensor based remote health monitoring,” IEEE Sensors J., vol. 12, no. 3, pp. 566-573, Mar. 2012.
[7]Girisha Garg, Vijander Singh, J.R.P. Gupta and A.P.Mittal, “Optimal Algorithm for ECG Denoising using Discrete Wavelet Transforms,” 2010 IEEE International Conference on Computational Intelligence and Computing Research, Coimbatore, India, Dec. 2010.
[8]Vincent, P., Larochelle, H., Bengio, Y. and Manzagol, P. A., “Extracting and composing robust features with denoising autoencoders,” Proc. 25th Int. Conf. Mach. Learn., pp. 1096-1103, Jul. 2008.
[9]Xiong, P., Wang, H., Liu, M. and Liu, X. “Denoising Autoencoder for Eletrocardiogram Signal Enhancement,” Journal of Medical Imaging and Health Informatics, vol. 5, pp.1804-1810, 2015.
[10]H.T. Chiang, Y.Y. Hsieh, S.W. Fu, K.H. Hung, Y. Tsao and S.Y. Chien, “Noise Reduction in ECG Signals Using Fully Convolutional Denoising Autoencoders,” IEEE Access, vol. 7, pp. 60806-60813, 2019.
[11]S. Hochreiter and J. Schmidhuber, “Long short-term memory,” Neural Comput., vol. 9, pp. 1735-1780, 1997.
[12]E. Dasan and I. Panneerselvam, “A novel dimensionality reduction approach for ECG signal via convolutional denoising autoencoder with LSTM,” Biomedical Signal Processing and Control, vol. 63, 2021.
[13]Jhang, Y. S., Wang, S. T., Sheu, M. H., Wang, S. H. and Lai, S. C., “Integration Design of Portable ECG Signal Acquisition with Deep-Learning Based Electrode Motion Artifact Removal on an Embedded System,” IEEE Access, vol. 10, pp. 57555-57564, 2022.
[14]K. He, X. Zhang, S. Ren and J. Sun, “Deep residual learning for image recognition”, Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), pp. 770-778, 2016.
[15]Jhang, Y. S., Wang, S. T., Sheu, M. H., Wang, S. H. and Lai, S. C., “Channel-Wise Average Pooling and 1D Pixel-Shuffle Denoising Autoencoder for Electrode Motion Artifact Removal in ECG,” Applied Sciences, vol. 12, no. 14:6957, 2022.
[16]W. Shi et al., “Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Network,” Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, vol. 2016-December, pp. 1874–1883, Sep. 2016.
[17]Vaswani, Ashish, et al., “Attention is all you need,” Advances in Neural Information Processing Systems 30(NeurIPS 2017), pp. 5998-6008, 2017.
[18]Doya, Kenji. "Bifurcations in the learning of recurrent neural networks," IEEE International Symposium on Circuits and Systems (ISCAS), vol. 6, pp. 2777-2780, 1992.
[19]Chen, M., Li, Y., Zhang, L., Liu, L., Han, B., Shi, W., & Wei, S., “Elimination of Random Mixed Noise in ECG using Convolutional Denoising Autoencoder with Transformer Encoder,” IEEE Journal of Biomedical and Health Informatics, vol. 28, no. 4, pp. 1993-2004, April 2024.
[20]L. F. Dong, Y. Z. Gan, X. L. Mao, Y. bin Yang, and C. Shen, “Learning Deep Representations Using Convolutional Auto-encoders with Symmetric Skip Connections,” ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings, vol. 2018-April, pp. 3006–3010, Nov. 2016.
[21]G. B. Moody, R. G. Mark and A. L. Goldberger, “Physionet: A Web-based resource for the study of physiologic signals”, IEEE Eng. Med. Biol. Mag., vol. 20, no. 3, pp. 70-75, 2001.
[22]M. George, M. Warren and M. Roger, “A noise stress test for arrhythmia detectors”, Comput. Cardiol., pp. 381-384, Nov. 1984.