基於強健性分散式訊源編碼的心電圖壓縮機制

Abstract

隨著生活水準的提升以及醫療科技的進步，高齡化社會是一個不可避免的未來趨 勢。因應城鄉差距及高齡化社會而興起的遠距醫療與健康照護，已成為先進國家重點發展的新興服務產業，也促成醫療電子產業轉向居家健康照護相關技術的開發。心電圖紀錄心臟搏動相關的電位變化資訊，常用於心血管疾病的診斷及治療。本論文旨在建立一適用於心電圖壓縮的分散式訊源編碼機制，其壓縮率及重建品質取決於訊源相關模型的建構及校驗子生成機制。我們首先針對MIT-BIH 心電圖資料庫，使用向量量化配合增益正規化的多重索引選擇機制，開發二位元交換演算法，據以建立最佳化訊源相關模型。並提出奇偶位元機制，透過擴展性奇偶位元查核矩陣衍生的柵狀碼圖作BCJR 演算法，能夠提高二位元通道的錯誤修正效能，藉以達到強健性分散式訊源編碼的心電圖壓縮機制。

In recent years, with the upgrading of the standard of living, as well as medical technology advances, wireless telemedicine and health-care has been one of the emerging research topics due to an aging society which is an inevitable trend. The electrocardiogram（ECG）records the changes of the heart beat-related potential, commonly used in the diagnosis and treatment of cardiovascular diseases. This thesis proposes a method for ECG compression based on distributed source coding. The compression ratio and the reconstruction quality depend on the source-related model, as well as syndrome former, therefore, we use gain-shaped vector quantization with multiple choice and binary switch algorithm to construct an optimized source-related model. Furthermore, by using trellis diagram formed from extended parity check matrix and BCJR decoding algorithm, we can improve the robustness of distributed source coding against binary symmetric channel errors.