- 學位論文
多導程心電與心音圖於早期診斷缺血型心臟病之應用
Early Detection of Ischemic Heart Disease Using Multi-lead ECG and Heart Sounds
摘要
近年來缺血型心臟病一直高居十大死因之首，在世界衛生組織的統計資料中，已開發國家中每一百位死亡人口就有約十六位是死於缺血型心臟病，可見其嚴重性。缺血型心臟病之致死率居高不下的主要原因，是由於其缺乏早期徵兆，且病患發生急性症狀到猝死的過程，往往是在很短的時間內，甚至有許多急性心肌梗塞的患者在猝死前完全沒有任何徵兆。因此，如何早期診斷並且把握治療的黃金一小時，成為近年來心臟病研究的重要課題。本研究使用非侵入式的十二導程心電圖與心音訊號應用在早期偵測缺血型心臟病，在急性冠心症發生之前偵測出心肌的缺血狀況，以期達到早期診斷的可能性，為病患爭取黃金治療時間。 本論文開發十二導程心電圖及心音訊號的監控系統及分析平台，以用於早期偵測缺血型心臟病。由於缺血型心臟病的患者，其心電圖的ST波與T波會短暫的升高或是反轉，本研究使用Support vector machine (SVM)與Sparse representation- based classification (SRC)兩種方法分析心電圖，以偵測心電圖的異常變化，此兩種方法與過去所提出的方法相比，能夠考量醫學知識中的心電圖異常，並配合使用有高準確率的機器學習方法，增加診斷的正確率。除了藉由心電圖早期診斷以外，本研究亦將非侵入式的心音訊號用於輔助診斷心肌功能的異常，本研究使用了適用於分析非線性及非穩態訊號的Hilbert-Huang Transform (HHT)方法，將心音訊號做可適性的時頻分析，以偵測其中與心肌功能相關的第三心音與第四心音。而在監控系統的部分，本論文開發的心電與心音圖監測系統，將多工方法應用在十二導程心電圖的記錄上，以符合無線傳輸和遠距照護的規格，並設計符合醫療需求的電子心音聽診裝置。 將本研究所開發之心電圖與心音分析平台與過去所提出之分析方法相比，本研究提出的方法其針對缺血型心臟病的診斷敏感度及整體準確率皆較高，而監控系統則有助於居家早期偵測缺血型心臟病，將此一整合型系統應用於遠距醫療照護，應可有效提升缺血型心臟病的早期診斷以及縮短就醫的時間。
並列摘要
Ischemic heart disease has become the first place of ten leading causes of death for many years. According to the statistic results from WHO, up to 16% of mortality is due to ischemic heart disease. The main reason of high death rate is its lack of early symptoms. Patients suffer from sudden death only after a short period of the occurring of acute coronary syndromes. Some even die without any early symptoms. Therefore, early detection of myocardial ischemia has become an important issue recently. In this study, we implemented a non-invasive 12-lead electrocardiogram (ECG) and a phonocardiogram (PCG) monitoring system, and high-accuracy analyzing methods are also proposed for the early detection of ischemic heart diseases. By the detection of the ischemia of cardiac muscles in its early stage, ischemic heart disease can be detected before the occurring of acute symptoms. Myocardial ischemia commonly manifests as ST- and T-wave changes on the ECG, or the third heart sound (S3) and the fourth heart sound (S4) of the PCG. For the analysis of ECG signals, we proposed two methods, support vector machine (SVM) and sparse representation-based classification (SRC), to detect abnormal ST-T complex. It integrates knowledge-based and novel classifying methods to extract essential information from ECG signals. In comparison with previous methods, the sensitivity for detecting myocardial ischemia is greatly improved using our methods. For the detection of S3 and S4, a time-frequency analysis method, Hilbert-Huang transform (HHT), was used to analyze non-linear and non-stationary PCG signals. This method can decompose the signal adaptively and acquire the instantaneous frequency. Therefore, all the abnormal components of PCG signals correlated to myocardial dysfunction can be detected simultaneously. The design of the monitoring of these non-invasive signals is based on remote home health care concepts. The recording of 12-lead ECG is designed using multiplexing technique suitable for wireless transmission. Moreover, the design of the electronic stethoscope is based on medical concepts with modulated equalizer. In this investigation, both analyzing methods and monitoring systems for 12-lead ECG and heart sound are proposed. The sensitivity and accuracy of the proposed methods are of better performance compared to previous methods. Furthermore, the whole monitoring system is aimed for remote home health care. With these concepts, detection of myocardial ischemia in its early stage using non-invasive home health care system could be feasible.