A new method, the orthonormal function model (OFM) is presented to identify the underlying features of electrocardiogram (ECG) waveforms. This method is based on the approximation properties of orthonormal functions. Among the set of all orthonormal functions, the Chebyshev polynomial has been selected, because it can uniformly approximate a broad class of functions and it gives the strongest convergence among all ultraspherical polynomials. Simulation results from normal and abnormal (ST depression) ECG waveforms indicate that the OFM has the following properties: (1) it has a residual error that decays to zero faster than for the linear predictor model (LPM): (2) it only needs a small model order for feature identification: (3) the model order for high resolution is smaller than for the LPM; and (4) most importantly the OFM can discriminate features from the ECG waveforms. The OFM can be successfully used for feature identification and as an aid in the classification and diagnosis of normal and abnormal patterns in the electrical activity of the heart.<<ETX>>
[1]
M.G. Bellanger,et al.
Digital processing of speech signals
,
1980,
Proceedings of the IEEE.
[2]
A. C. Young,et al.
Factor Analysis of the Electrocardiogram: Test of Electrocardiographic Theory Normal Hearts
,
1960,
Circulation research.
[3]
D A BRODY,et al.
Principal Factor Waveforms of the Thoracic QRS Complex
,
1964,
Circulation research.
[4]
Olle Pahlm,et al.
A Method for Evaluation of QRS Shape Features Using a Mathematical Model for the ECG
,
1981,
IEEE Transactions on Biomedical Engineering.
[5]
J. Fehlauer,et al.
Feature extraction by identification of a parameterized system model
,
1981
.
[6]
Kang-Ping Lin,et al.
QRS feature extraction using linear prediction
,
1989,
IEEE Transactions on Biomedical Engineering.