Electrocardiogram recognition using spectral correlation function

In this paper, a new representation method for electrocardiogram (ECG) signals analysis is suggested. This representation scheme is based on the spectral correlation function (SCF) which appears hidden periodicity of signals. The SCF presents a second-order statistical description in the frequency domain and can be used for several applications of ECG analysis such as classification. The SCF of each ECG signal is computed by using an efficient computational algorithm, called the FFT accumulation method (FAM). Then, the energy and standard deviations at different regions of bi-frequency plane of SCF are calculated as features. Afterwards, the features fed to a support vector machine (SVM) classifier for ECG classification. The obtained results indicate the efficiency of the new method in comparison with the previous studies.