利用二維卷積神經網路進行二維心電圖特徵之病徵分類及辨識

Classification and identification of symptoms using two-dimensional convolutional neural networks with two-dimensional electrocardiographic features

蔡宗穎(Tsung-Ying Tsai)

指導教授 ： 徐良育

中原大學/工學院/生物醫學工程研究所/碩士(2019年)

https://doi.org/10.6840/cycu201900809

摘要

心電圖(Electrocardiogram，ECG)是反映人體心臟健康的非常重要的工具，專家可以從ECG形態特徵中診斷出許多心臟異常之疾病。由於ECG於個體之間的差異及不可避免的雜訊，對專家於辨識及分類上為一大難題，因此需要較有效之方法以分類及辨識ECG數據中提取之特徵。
本研究提出一種利用12導程ECG之二維特徵圖同時分類多種病徵之自動分類方法，此方法中設計一個3層雙通道卷積神經網絡(Convolutional Neural Networks，CNN)，輸入則為12導程ECG訊號分別對應第2導程ECG訊號製作的二維特徵圖圖像。為了使CNN能夠更有效的學習ECG病徵，分別將原始ECG圖像分割為包含QRS複合波特徵之圖像及不包含QRS複合波特徵之圖像做為雙通道之輸入，並探討CNN架構中輸入圖像之大小與隱藏層層數數對於病徵分類準確度及時間之影響。
研究中使用開源PhysioBank MIT-BIH心律不整數據庫、聖彼得堡INCART12導程心律異常資料庫(St Petersburg INCART 12-lead Arrhythmia Database)及德國聯邦物理研究院心率資料庫(Physikalisch-Technische Bundesanstalt，PTB)，總共182名受測者來訓練及測試，並使用10層交叉驗證以驗證CNN分類之能力。在13種病徵之分類中，本系統分類平均準確度為84.92%、靈敏度為82.40%、特異性為82.79%，此結果表示對於心律不整之病徵已可達到有效之分類。由上述結果證明本系統可協助專家分析12導程ECG中之特徵，並加以分類其病徵。

關鍵字

心電圖 ； 12導程 ； 二維特徵圖 ； 卷積神經網路 ； 病徵分類

並列摘要

Electrocardiogram (ECG) is a very important tool to reflect the health of the human heart. Experts can diagnose many cardiac abnormalities from ECG morphological features. Due to the differences between individuals and unavoidable noises are the major problem for experts in identification and classification. Therefore, there is a need for powerful methods to classify and identify the information extracted from ECG datasets.
This study proposes an automatic classification method for simultaneously classifying multiple symptoms using a two-dimensional feature image of 12-lead ECG. In this method, a 3-layer two-channel convolutional neural network (CNN) is designed. The input are two-dimensional feature images that were pairing leadⅡ signal with all the other leads. To enable the CNN to learn ECG symptoms more effectively, the original ECG image was segmented into two images, one containing a QRS complex feature and an image does not including any QRS complex feature. Then the impact of the size of the input image and the number of hidden layers in the CNN architecture on the classification accuracy and time of the disease were explored.
In this study the open source PhysioBank MIT-BIH arrhythmia database, the St. Petersburg INCART 12-lead Arrhythmia Database and the Physikalisch-Technische Bundesanstalt (PTB) were used, a total of 182 subjects were included for training and testing. A 10-fold cross validation were used to verify the capabilities of the CNN classification. Among the 13 types of symptoms, the average accuracy of the system classification was 84.92%, the sensitivity was 82.40%, and the specificity was 82.79%. This result indicates that the classification of the symptoms of arrhythmia can be effective. It is also proved by the above results that the system can assist experts in analyzing the features of the 12-lead ECG and classifying the symptoms.

並列關鍵字

ECG ； 12 lead ； two-dimensional feature images ； convolutional neural network ； classification of symptoms

參考文獻

1.Benjamin, E.J., P. Muntner, A. Alonso, M.S. Bittencourt, C.W. Callaway, A.P. Carson, A.M. Chamberlain, A.R. Chang, S. Cheng, S.R. Das, F.N. Delling, L. Djousse, M.S.V. Elkind, J.F. Ferguson, M. Fornage, L.C. Jordan, S.S. Khan, B.M. Kissela, K.L. Knutson, T.W. Kwan, D.T. Lackland, T.T. Lewis, J.H. Lichtman, C.T. Longenecker, M.S. Loop, P.L. Lutsey, S.S. Martin, K. Matsushita, A.E. Moran, M.E. Mussolino, M. O'Flaherty, A. Pandey, A.M. Perak, W.D. Rosamond, G.A. Roth, U.K.A. Sampson, G.M. Satou, E.B. Schroeder, S.H. Shah, N.L. Spartano, A. Stokes, D.L. Tirschwell, C.W. Tsao, M.P. Turakhia, L.B. VanWagner, J.T. Wilkins, S.S. Wong, S.S. Virani, E. American Heart Association Council on, C. Prevention Statistics, and S. Stroke Statistics, Heart Disease and Stroke Statistics-2019 Update: A Report From the American Heart Association. Circulation, 2019. 139(10): p. e56-e528.

2.Xiang, Y., Z. Lin, and J. Meng, Automatic QRS complex detection using two-level convolutional neural network. Biomed Eng Online, 2018. 17(1): p. 13.

3.Thygesen, K., J.S. Alpert, A.S. Jaffe, M.L. Simoons, B.R. Chaitman, and H.D. White, Third universal definition of myocardial infarction. Circulation, 2012. 126(16): p. 2020-2035.

4.YILDIZ, O. and A. Arslan, Automated Auscultative Diagnosis System for Evaluation of Phonocardiogram Signals Associated with Heart Murmur Diseases. Gazi University Journal of Science. 31(1): p. 112-124.

5.Liu, S., Y. Wang, X. Yang, B. Lei, L. Liu, S.X. Li, D. Ni, and T. Wang, Deep Learning in Medical Ultrasound Analysis: A Review. Engineering, 2019. 5(2): p. 261-275.