Classification of ventricular tachycardia and fibrillation using fuzzy similarity-based approximate entropy

This paper presents an improved approximate entropy method for automatic diagnosis of ventricular fibrillation (VF) and ventricular tachycardia (VT). Approximate entropy (ApEn) is believed to provide quantitative information about the complexity of experimental data that are often corrupted with noise and short data length. However, the similarity definition of vectors is based on Heaviside function, of which the boundary is discontinuous and hard and may cause some problems in the validity and accuracy of ApEn. To overcome the problems ApEn encountered, an improved approximate entropy (iApEn) based on fuzzy membership function was proposed. Tests were conducted on independent, identically distributed (i.i.d.) uniform and Gaussian noises, chirp signal, MIX process, Rossler map, and Henon map. Compared with the standard ApEn, the iApEn showed better relative consistency and more robustness to noise when characterizing signals with different complexities. The proposed method was then applied to the VF and VT signals selected from MIT/BIH data sets. It is shown that, as a criteria for detecting between VF and VT, iApEn provides one with significantly higher (p=0.0017) accuracy rate (97.5%) than that of the standard ApEn method.