Shape characterization of atrial fibrillation using time-frequency analysis

Author

Stridh, M. ; Sörnmo, L.

Author_Institution

Dept. of Electroscience, Lund Univ., Sweden

fYear

2002

fDate

22-25 Sept. 2002

Firstpage

17

Lastpage

20

Abstract

A new method for characterization of atrial fibrillation in the surface ECG is presented which is based on a time-frequency distribution of the QRST-cancelled signal. With the present method, information regarding temporal variations in fibrillation frequency and waveform shape is extracted and investigated. The method is evaluated on a database including 182 recordings (obtained during resting conditions) from patients diagnosed with atrial fibrillation. The results show that a high fibrillation frequency is often associated with a sinusoidal shape whereas a low frequency is more associated with a sawtooth-like shape.

Keywords

electrocardiography; medical signal processing; QRST-cancelled signal; atrial fibrillation; database; fibrillation frequency; recordings; sawtooth-like shape; shape characterization; sinusoidal shape; surface ECG; temporal variations; time-frequency distribution; waveform shape; Atrial fibrillation; Data mining; Databases; Electrocardiography; Frequency estimation; Shape; Signal resolution; Spatiotemporal phenomena; Spectral analysis; Time frequency analysis;

fLanguage

English

Publisher

ieee

Conference_Titel

Computers in Cardiology, 2002

ISSN

0276-6547

Print_ISBN

0-7803-7735-4

Type

conf

DOI

10.1109/CIC.2002.1166696

Filename

1166696