本論文提出一一維之卷積神經網路(One-Dimension Convolutional Neural Network)分類模型，用以分類臨床心律不整與開放資料庫之心律不整疾病，此類神經網路模型之最大特色為其低維度且低層數的設計，以利於日後硬體化的實現。本論文之設計理念為利用類神經網路擁有特徵學習、特徵萃取等特性，讓模型自行做特徵抓取等動作，此法可有效省去傳統疾病辨識演算法的諸多前處理步驟，如波峰偵測、時頻分析、特徵萃取、特徵篩選等步驟，而這些步驟往往都是決定模型分類準確度的重要步驟，需要較多生醫訊號專業知識。在分類資料中加入了臨床資料以確保此模型能更貼近日常生活應用，且在分類模型上也盡量降低所需卷積層數讓計算複雜度降低並提升運算速度，此法將利於日後硬體的實現，對於攜帶裝置的可行性又大幅邁進。本論文所設計之分類模型在開放式資料庫中達到平均95.73%的分類準確度，而在臨床數據的分類上也達到了平均94.96%的準確度，相較於相關文獻，本論文所能硬體化之可能性較高，且準確度也相當足夠。特別地，本論文在建構訓練資料時，亦邀請專業的心臟科醫師進行一些心電圖的判讀教學，因此在建構訓練資料時有發現開放資料庫之標示略有不足的地方，在本文討論中會詳細提出，該發現有助於提升日後心律不整分類等研究之分類準確度。此模型應用於MIT資料庫及臨床資料(衛生署台南分部的IRB臨床資料)測試上皆有不
錯的表現，證明了此卷積神經網路的分類能力。

This thesis presents an arrhythmia classification system using one-dimensional (1D) convolutional neural network (CNN) and separated ECG segments to classify both the open-source MIT-BIH data and the clinical data from Tainan Hospital of Ministry of Health and Welfare. The CNN model is used to classify four common categories of arrhythmia as follows: Normal sinus rhythm, Atrial Fibrillation, Atrial Flutter and Fatal Arrhythmia (Ventricular Fibrillation and Torsade de points). The proposed 1D CNN model only needs a few amount of training data and neural network layers, but it still achieves an acceptable accuracy as well. We obtained an average accuracy of 95.73% by using the open-source MIT-BIHs’ data and achieved an average accuracy of 94.96% by using the clinical data. This work gets closer to reality because the separated segments are similar to the clinical diagnosis and is beneficial to integrate into the wearable devices in the future work due to its small training data and a few amount of neural network layers.

[1] 台灣老年人口比例 [Online].
Available : https://news.housefun.com.tw/mag/hf/14/article/12250070618.html
[2] WHO 公布之全球十大死因 [Online].
Available : http://www.who.int/mediacentre/factsheets/fs310/en/
[3] 神經元 [Online]
Available : http://www.hkpe.net/hkdsepe/human_body/neuron.htm
[4] 激勵函數 [Online].
Available : https://zhuanlan.zhihu.com/p/23906526
[5] 反向傳播 [Online].
Available : https://blog.csdn.net/taigw/article/details/50612963
[6] 卷積神經網路 [Online].
Available : http://www.how01.com/post_R19a5o9OYGdjK.html
[7] 卷積神經網路運作方式 [Online].
Available : https://medium.com/
[8] Zeiler M.D., Fergus R. (2014) Visualizing and Understanding Convolutional Networks. In: Fleet D., Pajdla T., Schiele B., Tuytelaars T. (eds) Computer Vision – ECCV 2014. ECCV 2014. Lecture Notes in Computer Science, vol 8689. Springer, Cham
[9] 心電圖 [Online].
Available : https://zh.wikipedia.org
[10] 導程介紹圖 [Online].
Available : https://smallcollation.blogspot.com/2013/07/electrocardiogram-ekgecg.html#gsc.tab=0
[11] 導程心電圖 [Online]
Available : https://smallcollation.blogspot.com/2013/07/electrocardiogram-ekgecg.html#gsc.tab=0
[12] MIT-BIH [Online]
Available : https://www.physionet.org/physiobank/database/mitdb/
[13] 心律不整 [Online]
Available : https://ecg-educator.blogspot.com/
[14] MIT資料庫檔案標記 [Online]
Available : https://www.physionet.org/cgibin/atm/ATM
[15] 正規化與收斂的關係 [Online]
Available : https://medium.com/
[16] U. R. Acharya, J. H. Tan, M. Adam. Automated Detection of Arrhythmias Using Different Intervals of Tachycardia ECG Segments with Convolutional. Information Sciences, 2017
[17] 一維卷積運算方式 [Online]
Available : https://zhuanlan.zhihu.com/p/34645443
[18] R. J. Martis, U. R. Acharya, H. Adeli, H. Prasad, J. H. Tan, K. C. Chua, C. L. Too, S. W. J. Yeo, L. Tong. Computer-aided diagnosis of atrial arrhythmia using dimensionality reduction methods on transform domain representation. Biomedical Signal Processing and Control 13: 295-305, 2014.
[19] R. J. Martis, U. R. Acharya, H. Adeli, H. Prasad, J. H. Tan, K. C. Chua, C. L. Too, S. W. J. Yeo, L. Tong. Computer-aided diagnosis of atrial arrhythmia using dimensionality reduction methods on transform domain representation. Biomedical Signal Processing and Control 13: 295-305, 2014.
[20] U. R. Acharya, H. Fujita, M. Adam, S. L. Oh, J. H. Tan, V. K. Sudarshan, J. E. W. Koh. Automated characterization of Arrhythmias using nonlinear features from tachycardia ECG beats. IEEE International Conference on Systems, Man, and Cybernetics, 2016.
[21] U. Desai, R. J. Martis, U. R. Acharya, C. G. Nayak, G. Seshikala, S. K. Ranjan. Diagnosis of multiclass tachycardia beats using recurrence quantification analysis and ensemble classifiers. Journal of Mechanics in Medicine and Biology, 16(1), 2016.