摘要
12導程心電圖訊號特徵點辨識，是建立智慧型心電圖自動診斷系統的重要關鍵技術。本研究所提出的演算法是修改公開程式碼“ECGPUWAVE”與元智大醫療資訊暨遠距醫學實驗室所開發的心電圖訊號處理技術，設計構思主要是基於訊號的斜率、振幅、以及波形寬度的分析。在本研究中，蒐集了205筆臨床12導程心電圖檔案，包含心房顫動、急性心肌梗塞、以及正常的病例。結果顯示(1)R波辨識的靈敏度平均為99.7%；(2)Q波與S點的靈敏度平均分別為99.51%和99.55%；(3)T波的靈敏度平均為97.96%；(4)P波的靈敏度可高達98.49%。傳統的電腦心電圖自動診斷準確率不高，而所提出的演算法能有效地辨識出臨床心電圖訊號中的特徵點且準確率較高。因此，在未來能以此演算法發展出一套自動化智慧型心電圖診斷系統。
並列摘要
It is crucial to develop an algorithm that can identify the characteristic points of 12 lead ECG , such as P, QRS, and T waves, for the development of computer-assisted 12-lead ECG interpretor . The proposed algorithm is modified from “ECGPUWAVE”, which is an open-source software, and YZU medical informatics and telemedicine lab developed ECG signal processing technique. The ECG characteristic points are detected by the slope, amplitude, and width of the first derivative of ECG waveforms. In this study, 205 clinical 12-lead ECGs, which are confirmed by cardiologists as atrial fibrillation, acute myocardial infarction, and normal cases, are collected. The results show that (1) the sensitivity of R wave detection is 99.7%; (2) the sensitivities of Q-wave and S-point are 99.51% and 99.55% , respectively ; (3)the sensitivity of T-wave is 97.96%; and (4) the sensitivity of P wave can be up to 98.49%. In conclusions, this developed algorithm can effectively detect the characteristic points of clinical ECG signals and thus can be developed as ECG interpretor in the near future.