Session S52.2

Segmentation of Heart Sound Recordings by a Duration Dependent Hidden-Markov Model

SE Schmidt*, E Toft, C Holst-Hansen, C Graff, JJ Struijk

Aalborg University
Aalborg, Denmark

Digital stethoscopes offers new opportunities for computerized analysis of heart sounds. Segmentation of hearts sounds into periods related to the first and second heart sound (S1 and S2) is fundamental in the analysing process. However segmentation of heart sounds recorded with handhold stethoscopes in clinical environments is often complicated by recording and background noise.
A Duration-dependent Hidden Markov Model (DHMM) is proposed for robust segmentation of heart sounds. The DHMM identifies the most likely sequence of physiological heart sounds, based on duration of the events, the amplitude of the signal envelope and a predefined model structure. The DHMM model was developed and tested with heart sounds recorded bedside with a commercially available hand held stethoscope from a selected population of patients referred for coronary angiography. The method was initially trained on 40 patients and healthy subjects.
The DHMM identified 739 S1 and S2 sounds out of 744 which corresponds to 99.3% (CI: 98.4-99.7%) sensitivity in 60 test- patients and misplaced 7 sounds out of 746 identified sounds which corresponds to 99.1% (CI: 98.1-99.5%) positive predictivity. These results indicate that DHMM is an appropriate model of the heart cycle and suitable for segmentation of clinically recorded heart sounds.

(Abstract Control Number: 203)