Automatic pediatric congenital heart disease classification based on heart sound signal
作者：
Highlights：
• A database of heart sounds in CHD is established using electronic stethoscopes and is proposed to be made publicly available online.
• We propose a segments-based segmentation algorithm for cardiac cycle identification of heart sound signals.
• Effective time and frequency domain features from different cardiac stages are designed for classifier training and testing.
• We establish a Random Forest and Adaboost based majority voting classifier for pediatric CHD classification.
• Through extensive experiments and comparisons, the proposed algorithms are verified to be competitive.
摘要
•A database of heart sounds in CHD is established using electronic stethoscopes and is proposed to be made publicly available online.•We propose a segments-based segmentation algorithm for cardiac cycle identification of heart sound signals.•Effective time and frequency domain features from different cardiac stages are designed for classifier training and testing.•We establish a Random Forest and Adaboost based majority voting classifier for pediatric CHD classification.•Through extensive experiments and comparisons, the proposed algorithms are verified to be competitive.
论文关键词：Pediatric congenital heart diseases,Heart sound,Feature extraction,Segmentation,Classification
论文评审过程：Received 21 June 2021, Revised 31 December 2021, Accepted 15 February 2022, Available online 19 February 2022, Version of Record 21 February 2022.
论文官网地址：https://doi.org/10.1016/j.artmed.2022.102257
