AdvFPCG-Delineator: Advanced delineator for fetal phonocardiography. (August 2020)

• Record Type: Journal Article
• Authors: Tomassini, Selene; Sbrollini, Agnese; Strazza, Annachiara; Sameni, Reza; Marcantoni, Ilaria; Morettini, Micaela; Burattini, Laura
• Is Part Of: Biomedical signal processing and control. Volume 61(2020)
• Publication Date: 2020-08
• Subjects: Fetal heart-rate extraction -- Fetal heart-sound identification -- Fetal phonocardiography -- Phonocardiographic features -- Scalogram -- Wavelet denoising; Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2020.102021
• Languages: English
• ISSNs: 1746-8094

Abstract

Graphical abstract: Highlights: The major fetal heart sounds, S1 and S2, belong to fetal phonocardiogram (FPCG). This paper proposes AdvFPCG-Delineator as tool to identify fetal S1 and S2. S1 and S2 are identified by a threshold-based algorithm on FPCG scalogram. The algorithm was tested on 119 FPCG and compared to simultaneous cardiotocograms. AdvFPCG-Delineator proved to be a reliable method to identify S1 and S2. Abstract: Fetal phonocardiogram (FPCG) consists in the recording of fetal heart sounds by means of a sensor placed on the mother's abdominal surface. Usually, FPCG includes two major sounds for each fetal cardiac cycle: S1, produced by the sudden closure of mitral and tricuspid valves, and S2 produced by the closure of aortic and pulmonary valves. The aim of the present study was to propose AdvFPCG-Delineator for automatic fetal S1 and S2 identification and to demonstrate its reliability in different clinical conditions. The method consists of a wavelet-based filtering procedure followed by the computation of the scalogram, from which S1 and S2 were identified using a threshold-based algorithm. AdvFPCG-Delineator was tested on the "Simulated Fetal PCGs database" (37 FPCG signals) and on the experimental "Shiraz University fetal heart sounds database" (119 FPCG signals), both available at PhysioNet (https://physionet.org ). Manual S1 and S2 annotations and simultaneously acquired cardiotocographic recordings were used to compute reference fetal heart rate (FHR) for the simulated and experimental databases, respectively. No statistically significant difference was observed between estimated vs reference FHR (140 bpm vs 140 bpm, respectively) for the simulated database, for which AdvFPCG-Delineator was also able to track beat-to-beat variability (correlation over 92%). Additionally, no statistically significant difference was observed between estimated vs reference FHR (141 bpm vs 140 bpm, respectively) for the experimental database, even when stratifying by clinical conditions (maternal age, gestational age, etc.). In conclusion, AdvFPCG-Delineator proved to be a reliable method to automatically identify S1 and S2 from fetal phonocardiograms.