Complete Ensemble EMD and Hilbert transform for heart beat detection

MARCELO ALEJANDRO COLOMINAS; GASTÓN SCHLOTTHAUER; MARÍA EUGENIA TORRES

Paraná

Congreso; VI Congreso Latinoamericano de Ingeniería Biomédica; 2014

Consejo Regional de Ingeniería Biomédica para América Latina - Sociedad Argentina de Bioingeniería - Facultad de Ingeniería (UNER)

The problem of heart beat detection is addressed with a new noise-assisted Empirical Mode Decomposition (EMD) method. Although existing algorithms show very good performances in good conditions, they have some difficulties in the presence of complicated waveforms, high levels of noise and strong baseline drift. Here we present a fully automated and unsupervised algorithm. Hilbert transform is used for the construction of envelopes of the extracted components and an entropy-like criterion is used for mode selection. The more difficult electrocardiogram (ECG) derivations are used in this work and the results are compared with the algorithm of Afonso et al., showing a better performance for our algorithm.