在心室顫動狀況下利用訊號分析準確預測除顫時機

Abstract

在這篇論文中，我們提出了一個新的演算法來預測在心室顫動狀況下預測除顫是否成功。實驗結果顯示我們的演算法比傳統的方法AMSA 和DFA 有更好的預測效果。此外，我們結合了我們的演算法與AMSA 這個方法來進行預測，實驗結果顯示結合兩種做法的結果的檢測率可高達0.933 並控制誤警率在0.507，結果比只用AMSA 一個預測方法的檢測率為0.767 與誤警率為0.453 效果來的好許多。再者，我們利用兩個理論來分析我們提出的演算法，理論的模擬結果與我們實驗的結果非常相近。因此，分析的結果可以用來評估我們的提出的演算法。

In this thesis, we propose a new algorithm to predict whether a defibrillation of a ventricular fibrillation ECG record is success or not. Experimental results show that the proposed algorithm outperforms conventional methods including amplitude spectrum area (AMSA) and detrended fluctuation analysis (DFA). In addition, when the proposed method combines with the AMSA method, the detection rate can be up to 0.933 with a false alarm rate 0.507, which greatly outperform the AMSA alone method that has a detection rate 0.767 with a false alarm rate 0.453. Moreover, we proposed two approximations to analyze the performance of the proposed algorithm. The theoretical results and simulation results have a satisfactory agreement. Hence, the analytical results can be used to evaluate the proposed algorithm.