Usefulness for Time-Frequency Analysis of P Wave to Predict Paroxysmal Atrial Fibrillation

Abstract

Purposes: Paroxysmal atrial fibrillation (PAF) is known to be a potential risk of systemic thromboembolism, but the prediction of PAF is difficult. In this study, the frequency distribution within P wave in patients with and without PAF. Methods: ECG (leads V1 and II) were recorded from patients with and without PAF-patient (n=20), and saved as MFER file. After QRS-T waves were subtracted from ECG, time-frequency analysis using wavelet transform was performed as following conditions: frequency range; 1–200 Hz, virtual sampling rate; 1 KHz, analyzing wavelet; Gabor. The wavelet-transformed ECG signals were integrated during 120 msec from the onset of P wave (ITFP: integrated time frequency power). The ITFPs of sequential 10 P waves was summed (S-ITFP). S-ITFPs in 80 frequency bands plotted against frequency axis indicate the spectral profile of P wave. Results: The spectral profile of P waves In lead II was different between with and without PAF-patients. The frequency power was greater in lead V1 than lead II, ranged from 20–30 Hz. Conclusions: It was suggested that the frequency profile of P wave is beneficial for the prediction of PAF-patient.