TY - JOUR
T1 - Event synchronous adaptive filter based atrial activity estimation in single-lead atrial fibrillation electrocardiograms
AU - Lee, Jeon
AU - Song, Mi Hye
AU - Shin, Dong Gu
AU - Lee, Kyoung Joung
N1 - Funding Information:
Acknowledgments This work was supported by the Technology Innovation Program (10040408) funded by the Ministry of Knowledge Economy (MKE, Korea).
PY - 2012/8
Y1 - 2012/8
N2 - In this paper, an event synchronous adaptive filter (ESAF) is proposed to estimate atrial activity (AA) from a single-lead AF ECG in real time. The proposed ESAF is a kind of adaptive filter designed to have the reference fed with the impulse train synchronized with the R peak in a raw atrial fibrillation (AF) ECG and to input the timely delayed AF ECG into the primary input. To assess the performance, for ten simulated AF ECGs, the crosscorrelation coefficient (ρ) and the normalized mean square error (NMSE) between estimated AAs and ten original simulated AAs were calculated and, for ten real AF ECGs, the ventricular residue (VR) in QRS interval and similarity (S) in non-QRS interval were computed. As a result, these four parameters were revealed as ρ = 0.938 ± 0.016 and NMSE = 0.243 ± 0.051 for simulated AF ECGs and VR = 1.190 ± 0.476 and S = 0.967 ± 0.041 for real AF ECGs. These results were found to be better than those of the averaged beat subtraction (ABS) method, which had been previously considered the only way to estimate AA automatically in real time. In conclusion, even with singlelead AF ECGs, the proposed method estimated AAs accurately and calculated the atrial fibrillatory frequencies, the most valuable index in AF maintenance and therapy evaluation, with a remarkably low computational cost.
AB - In this paper, an event synchronous adaptive filter (ESAF) is proposed to estimate atrial activity (AA) from a single-lead AF ECG in real time. The proposed ESAF is a kind of adaptive filter designed to have the reference fed with the impulse train synchronized with the R peak in a raw atrial fibrillation (AF) ECG and to input the timely delayed AF ECG into the primary input. To assess the performance, for ten simulated AF ECGs, the crosscorrelation coefficient (ρ) and the normalized mean square error (NMSE) between estimated AAs and ten original simulated AAs were calculated and, for ten real AF ECGs, the ventricular residue (VR) in QRS interval and similarity (S) in non-QRS interval were computed. As a result, these four parameters were revealed as ρ = 0.938 ± 0.016 and NMSE = 0.243 ± 0.051 for simulated AF ECGs and VR = 1.190 ± 0.476 and S = 0.967 ± 0.041 for real AF ECGs. These results were found to be better than those of the averaged beat subtraction (ABS) method, which had been previously considered the only way to estimate AA automatically in real time. In conclusion, even with singlelead AF ECGs, the proposed method estimated AAs accurately and calculated the atrial fibrillatory frequencies, the most valuable index in AF maintenance and therapy evaluation, with a remarkably low computational cost.
KW - Atrial activity
KW - Averaged beat subtraction
KW - Electrocardiogram
KW - Event synchronous adaptive filter
KW - Fibrillatory rate
UR - http://www.scopus.com/inward/record.url?scp=84864353471&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84864353471&partnerID=8YFLogxK
U2 - 10.1007/s11517-012-0931-7
DO - 10.1007/s11517-012-0931-7
M3 - Article
C2 - 22718318
AN - SCOPUS:84864353471
SN - 0140-0118
VL - 50
SP - 801
EP - 811
JO - Medical and Biological Engineering and Computing
JF - Medical and Biological Engineering and Computing
IS - 8
ER -