第1筆 /總和 1 筆
/ 1 筆
頁籤選單縮合
題名 | Ecg Analysis and Processing using Wavelets and Other Methods |
---|---|
作者 | Celler,Branko,; Chung,Grace Yuet Chee; Phillips,Chris; | 書刊名 | 醫學工程 |
卷期 | 9:2 1997.04[民86.04] |
頁次 | 頁8-17 |
分類號 | 415.3023 |
關鍵詞 | ECG; Ventricular tachycardia; Wavelets analysis; Ventricular late potentials; |
語文 | 英文(English) |
英文摘要 | Detection of ventricular late potentials is well established as a non- invasive tool for identification of patients at risk for ventricular tachycardia. The conventional methodology for detecting late potentials is time-domain analysis of the high-resolution signal averaged ECGs. Time-domain analysis however, as a predictor of vulnerability to ventricular techycardia, suffers from lack of reliability. Predictive values range between 20% to 95% with an average of 64%. In this paper we compare the results obtained from Wavelet Transform analysis using the Morlet wavelet, with results obtained using both conventional time-domain analysis and time frequency analysis based on measures of spectral turbulence. These method are applied to ECG data collected from three groups. Normals were students in the age group of 18-25 years of age with no known history of heart disease, abnormals were either in a post mild myocardial infarction (MI) recovery phase or in an acute phase of MI. Significant and consistent differences were observed in the wavelet transforms of normal and acute MI subjects Normals exhibited a highly regular and smooth time frequency surface with energy contributions concentrated in a narrow QRS complex. Post MI subjects demonstrated abrupt and irregular pockets of localised energy interspersed throughout the time frequency plane, and extending beyond the QRS complex. Significant statistical differences were found between normal and acute MI subjects but not between normals and the post mild MI subjects. Results using Simson's method with those based on spectral turbulence were similarly in poor agreement. Quantitative methods for the normalisation and comparison of energy dispersion require optimisation and more sophisticated methods based on discriminant analysis may prove more sensitive. Time frequency analysis using other wavelets such as Daubechies or the Mexican Hat, and alternative transforms such as the Wigner-Ville transforms need to be rigorously examined and compared statistically for sensitivity and effectiveness. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。