- 學位論文
以生成對抗神經網路實現心房顫動心電圖合成訊號之研究
The Study of Synthesizing The Atrial Fibrillation ECG Signal Using Generative Adversarial Network
摘要
心房顫動是一種心房異常快速跳動的心律不整。根據流行率調查，約百分之二的人曾經有發生過心房顫動，且隨著年齡的增長，心房顫動發生的機率也相對地上升。引起心房顫動的原因，普遍被認為是心房組織快速的不正常放電，導致患者常形容心跳忽快忽慢而不規律的跳動。 近年來，隨著深度學習的崛起，透過大量的資料來訓練類神經網路，電腦輔助診斷變得越來越重要，此技術在臨床漸漸成為醫生診斷的重要輔助手段。在過去的醫療診斷中，心房顫動的參考樣本數少，但卻又對人體健康有舉足輕重的影響。 生成式對抗神經網路是生成模型的一種，因其能夠生成逼真且高質量資料而備受關注。生成式對抗神經網路可以複製所有的分佈資料，並在圖像、節奏、音樂甚至語音等不同領域創建對資料集的模仿。本研究的主要目的是利用真實心房顫動的心電圖資料使用神經網路來合成更多的各式各樣的心房顫動心電圖，實驗利用生成式對抗神經網路的深度學習模型，對當前的資料數進行訓練，探討真實訊號與合成訊號的差別與效果。
並列摘要
Atrial fibrillation is an arrhythmia in which the atrium beats abnormally quickly. According to the prevalence survey, about 2% of people have experienced atrial fibrillation, and with age, the probability of atrial fibrillation increases relatively. The cause of atrial fibrillation is generally considered to be the rapid and abnormal discharge of the atrial tissue, causing patients to describe the heartbeat as sudden and slow and irregular. In recent years, with the rise of deep learning, computer-aided diagnosis has become more and more important through a large amount of data to train neural networks. This technology has gradually become an important aid for doctors in clinical diagnosis. In the past medical diagnosis, the number of reference database for atrial fibrillation was small, but it had a significant impact on human health. Generative adversarial network is a kind of generative model, which has attracted much attention because of its ability to generate realistic and high-quality data. Generative adversarial networks can learn all distributed data and create imitations of data sets in different fields such as image, rhythm, music, and even voice. The main purpose of this research is to use the electrocardiogram data of real atrial fibrillation to synthesize more various atrial fibrillation electrocardiograms using Generative adversarial networks. Discussing the difference and effects of real and synthesized signals.
參考文獻
延伸閱讀
- 高東韋（2021）。以單導程心電圖訊號進行心房顫動偵測之研究〔碩士論文，國立臺灣大學〕。華藝線上圖書館。https://doi.org/10.6342/NTU202101035
- 蔡柏鋐（2016）。基於心電圖辨識心室頻脈與心室顫動及其用於預測心因性猝死之研究〔碩士論文，國立中正大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0033-2110201614061301
- Li, Y. L. (2013). 應用於生醫系統之心律不整偵測與心電信號壓縮電路 [master's thesis, National Taiwan University]. Airiti Library. https://doi.org/10.6342/NTU.2013.10407
- 朱湘麟（2006）。利用拉普拉斯電位分析心房纖維性顫動之停止機制〔碩士論文，中原大學〕。華藝線上圖書館。https://doi.org/10.6840/CYCU.2006.00466
- Kao, T., Su, Y. Y., Lu, C. C., Tai, C. T., Chen, S. A., Lin, Y. C., & Tso, H. W. (2005). Differentiation of Atrial Flutter and Atrial Fibrillation from Surface Electrocardiogram Using Nonlinear Analysis. Journal of Medical and Biological Engineering, 25(3), 117-122. https://www.airitilibrary.com/Article/Detail?DocID=16090985-200509-25-3-117-122-a