Adaptive SVR Denoising Algorithm for Fetal Monitoring System

Abstract

This paper proposes a new fetal monitoring system, including the acquisition and the processing of fetal heart sounds (FHS). Based on the foundation of the stethoscope principle, a single-channel, non-invasive sensor is designed to acquire the fetal heart sounds, in which polyvinylidene fluoride (PVDF) membrane material is used as the core transducer. In the fetal heart sounds processing part, we propose a new method for denoising based on adaptive support vector regression (SVR) which has a good performance on curve fitting and effectively weakens the interference of additive noise. Thus, the clean fetal heart signals extracted from the interfered source can be further utilized to draw the fetal phonocardiogram (FPCG) and calculate the fetal heart rate (FHR).