Title :
Improved frequency-domain analysis of ventricular late potentials
Author_Institution :
Dept. of Electron. Eng., Lunghwa Univ. of Sci. & Technol.
Abstract :
Many methods of segment selection of various lengths and locations in the high-frequency spectral analysis of ventricular late potentials (VLPs) in signal-averaged electrocardiograms (SAECGs) have been demonstrated to yield conflicting results in clinical applications. This work compares of a new locally developed method with two existing methods for detecting VLPs in the frequency domain. A total of 154 normal individuals, 94 patients with frequent ventricular premature contraction (VPC) and 26 patients with sustained ventricular tachycardia (VT) were recruited for the study. Two existing methods use a 120 ms time segment, starting from 20 ms before spatial vector velocity <5 mV/s and vector magnitude <40 muV at the terminal QRS complex, to analyze the VLPs. A locally developed 80 ms segment, starting from 60 ms before the QRS offset, was shown to outperform these two currently adopted methods. The areas under the receiver operating curves (AUCs) of the root-mean-square amplitude (RMSA) in the 60 to 120 Hz band were 82.2% versus 60.2% and 55.2%, and the AUCs of the RMSA ratio (RMSAR) (100times[60 to 120 Hz RMSA/0 to 120 Hz RMSA]) were 77.2% versus 54.9% and 53.1%. The locally developed method, which uses the QRS offset as a reference for selecting segments is determined substantially to improve VLPs analysis in the frequency domain
Keywords :
bioelectric potentials; electrocardiography; frequency-domain analysis; medical signal detection; medical signal processing; sensitivity analysis; spectral analysis; 120 ms; 60 to 120 Hz; 80 ms; QRS signal detection; area under receiver operating curves; frequency-domain analysis; high-frequency spectral analysis; normal individuals; root-mean-square amplitude; signal-averaged electrocardiograms; sustained ventricular tachycardia patients; ventricular late potential detection; ventricular premature contraction patients; Aging; Cardiology; Electrocardiography; Frequency domain analysis; Heart rate variability; Hospitals; Signal analysis; Signal resolution; Spectral analysis; Time domain analysis;
Conference_Titel :
Computers in Cardiology, 2005
Conference_Location :
Lyon
Print_ISBN :
0-7803-9337-6
DOI :
10.1109/CIC.2005.1588141