Detection of the S2 split using the Hilbert and Wavelet transforms



Pulmonary hypertension (PH) is a serious heart condition that is dicult
to diagnose in ambulatory settings. Heart sounds is one of the most relevant diagnosis
signal in this context. Usually, PH leads to wide S2 split between the aortic valve close
sound (A2) and the pulmonary valve close sound (P2). Even though the progresses made,
there isn't still an automatic and non-supervised way to eectively measure the S2 Split.
Results based on the wavelet transform, Wigner-Ville distribution and Chirp models are
presented in the literature. However, most methods require human intervention and de-
pend on parameters with high intra and inter-patient variability, showing statistically low
signicant results. Furthermore, most of these methods are evaluated in synthesized sig-
nals or non-human heart sounds. In this paper three approaches for S2 Split detection are
presented and compared based on the onset of the A2 and P2 components. The approaches
are based on the Hilbert transform's instantaneous amplitude (IA) and frequency (IF), and
on the continuous wavelet transform IF (CWT). Results from a 226 manually annotated
S2 sound database will be presented and discussed. The best approach, a combination of
the Hilbert transform IA and IF, obtained an average error of 3:4 2:8ms (correlation
= 0:99) on the detection of the A2 onset, and an error of 10:3 11:2ms (correlation
= 0:99) on the detection of the P2 start instant.




medical informatics

Related Project



Congresso de Métodos Numéricos em Engenharia – CMNE 2011., January 2011

PDF File

Cited by

Year 2013 : 1 citations

 Abdelghani Djebbari, Fethi Bereksi-Reguig (2013). “Detection of the valvular split within the second heart sound using the reassigned smoothed pseudo Wigner-Ville distribution”, BioMedical Engineering OnLine 2013, 12:37 doi:10.1186/1475-925X-12-37

Year 2011 : 1 citations

 1. Hamza Cherif, L., and S. M. Debbal. "Algorithm for detection of the internal components of the heart sounds and their split using a Hilbert transform." Journal of medical engineering & technology 37.3 (2013): 220-230.