Adaptive Differential Denoising for Respiratory Sounds Classification

TL;DR

This paper introduces an Adaptive Differential Denoising network that enhances respiratory sound classification by effectively suppressing noise while preserving diagnostic features, leading to improved accuracy on a standard dataset.

Contribution

The paper presents a novel adaptive denoising approach combining spectral masking, differential attention, and joint optimization, which outperforms existing methods in respiratory sound classification.

Findings

Achieved 65.53% Score on ICBHI2017 dataset.

Improved performance by 1.99% over previous state-of-the-art.

Demonstrated effectiveness of adaptive denoising in noisy respiratory sound analysis.

Abstract

Automated respiratory sound classification faces practical challenges from background noise and insufficient denoising in existing systems. We propose Adaptive Differential Denoising network, that integrates noise suppression and pathological feature preservation via three innovations: 1) Adaptive Frequency Filter with learnable spectral masks and soft shrink to eliminate noise while retaining diagnostic high-frequency components; 2) A Differential Denoise Layer using differential attention to reduce noise-induced variations through augmented sample comparisons; 3) A bias denoising loss jointly optimizing classification and robustness without clean labels. Experiments on the ICBHI2017 dataset show that our method achieves 65.53\% of the Score, which is improved by 1.99\% over the previous sota method. The code is available in https://github.com/deegy666/ADD-RSC