Cardiac anomaly detection considering an additive noise and convolutional distortion model of heart sound recordings

作者：

Highlights：

• Heart sounds can be corrupted with additive noise and convolutional degradation (sensor effect).

• We mathematically analyze the effect of additive noise & convolutional distortion on heart sounds.

• We propose a fusion of linear and logarithmic features input to a CNN for disease classification.

• Achieved superior results compared to existing methods in presence of sensor/noise variability.

摘要

•Heart sounds can be corrupted with additive noise and convolutional degradation (sensor effect).•We mathematically analyze the effect of additive noise & convolutional distortion on heart sounds.•We propose a fusion of linear and logarithmic features input to a CNN for disease classification.•Achieved superior results compared to existing methods in presence of sensor/noise variability.

论文关键词：Additive noise and convolutional distortion,Stethoscope variability,Heart sound classification

论文评审过程：Received 29 October 2021, Revised 17 September 2022, Accepted 2 October 2022, Available online 7 October 2022, Version of Record 15 October 2022.

论文官网地址：https://doi.org/10.1016/j.artmed.2022.102417