即時心電圖分析系統研製及消趨勢分析於心衰診斷之應用

Realization of Real-Time Electrocardiogram Systems and Their Applications to Diagnosis of Patients with Congestive Heart Failure

張彬(Chang, Ben)

指導教授 ： 鐘太郎

國立清華大學/電機資訊學院/電機工程學系/博士(2010年)

https://doi.org/10.6843/NTHU.2010.00633

摘要

摘要
充血性心衰竭(congestive heart failure, CHF)是各種心血管疾病的終末階段，主要特徵為心肌收縮力減弱，心排血量減少，因而不能滿足機體組織細胞代謝的需要，同時靜脈回流受阻，靜脈系統淤血，從而出現以組織血液灌注不足，以及肺循環和體循環淤血等一種複雜而嚴重的臨床綜合症狀，它具有高發病率和高病死率的二高特點，尤其是在狀況「輕」的情況下，心衰竭往往由於沒有共同承認的定義，以及難以診斷出來。即便是使用最好的治療，心衰竭的年死亡率依舊高達10%。而心衰竭亦是導致65歲以上的老年人入院的主要原因。近年心衰的發病率預料仍將繼續上升，有可能成為本世紀最嚴重的心血管病症，因此心臟系統疾病的正確診斷是當今醫學界面臨的最為迫切需解決的問題。本論文的研究重點包括心率變異(heart rate variability, HRV)的各式即時分析器(real-time analyzer, RTA)及其於CHF診斷之應用。主要研製無線式心電圖(electrocardiogram, ECG)量測系統、基於個人電腦(Personal computer, PC)的虛擬RTA，並將系統實際應用於CHF的診斷。本文首先說明ECG的研製，核心採用自製的低成本電路，來取代昂貴之儀器放大和通信晶片，大大改善傳統有線不便問題，並克服雜訊影響。其次，說明RTA的研製，通過虛擬架構，並修改PC之電路功能，研製出分別具有濾波、間隔期檢測、小波分析、和消趨勢分析(detrended fluctuation analysis, DFA)等功能的各式RTA。接著，採用DFA、近似熵及複雜度分析等三個特徵，給可適應性神經模糊推理系統診斷CHF。最後使用支持向量機(support vector machine, SVM)，探討DFA在接近一日觀察HRV中之應用，並總結出7PM ~9PM是診斷CHF的最佳時機。

關鍵字

心率變異 ； 心臟衰竭 ； 消趨勢分析

並列摘要

Abstract

In this dissertation, the realization of real-time analyzer (RTA) for heart rate variability (HRV) analysis and their applications to congestive heart failure (CHF) are investigated. Implementations include wireless electrocardiogram (ECG) and personal computer (PC) based virtual RTAs, and their applications to CHF are studied. Firstly, we present the realization of ECG. Using low-cost components to replace expensive instrumentation and communication ICs in the circuits, we can not only reduce the system cost, but also resolve the cable inconvenience problems of the data acquisition. In addition, our result is good in noise immune. Next, through the virtual architecture, and modify the circuits of PC, we realized various RTAs capable of filtering, interval detecting, wavelet analyzing, and detrended fluctuation analysis (DFA), respectively. Then, use of DFA, approximate entropy and complexity measure as features of the adaptive neuro-fuzzy inference system (ANFIS) to CHF diagnosis is described. Finally, based on the support vector machine (SVM) we present the application of DFA during circadian observation, and conclude that 7PM~9AM is the best timing to diagnose CHF patient via the α1 scaling exponent of DFA.

並列關鍵字

heart rate variability ； congestive heart failure ； detrended fluctuation analysis

參考文獻

44Vapnik, V., and A. Lerner, Pattern recognition using generalized portrait method. Automat. Remote Contr, 24:774-780, 1963.

1Allegra, J. R., D. G. Cochrane, and R. Biglow. Monthly, weekly, and daily patterns in the incidence of congestive heart failure. Acad. Emerg. Med. 8(6):682-685, 2001.

2Aschoff, J., S. Daan, and G. A. Groos. Vertebrate circadian systems structure and physiology. New York: Springer-Verlag, 1982.

3Bär, K. J., M. Koschke, S. Berger, S. Schulz, M. Tancer, A. Voss, and V. K. Yeragani. Influence of olanzapine on QT variability and complexity measures of heart rate in patients with schizophrenia. J. Clin. Psychopharmacol. 28(6):694-698, 2008.

4Barash, D., R. A. Silverman, P. Gennis, et al. Circadian variation in the frequency of myocardial infarction and death associated with acute pulmonary edema. J. Emerg. Med. 7:119-121, 1989.