Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

XAI-Driven Spectral Analysis of Cough Sounds for Respiratory Disease Characterization

Patricia Amado-Caballero, Luis Miguel San-José-Revuelta, María Dolores Aguilar-García, José Ramón Garmendia-Leiza, Carlos Alberola-López, Pablo Casaseca-de-la-Higuera

TL;DR

The paper addresses the challenge of disease differentiation from cough sounds by introducing an eXplainable AI (XAI) workflow that uses occlusion maps on CNN-derived cough spectrograms to weight spectral information. By extracting spectral features from these weighted spectrograms, the study demonstrates significant differences between disease groups, most notably COPD, which shows greater spectral variability than other pathologies. In contrast, analyses on unweighted, raw spectrograms fail to reveal such distinctions, highlighting the value of XAI-guided feature extraction. The approach improves interpretability and has potential to enhance non-invasive, cough-based diagnostics for respiratory diseases.

Abstract

This paper proposes an eXplainable Artificial Intelligence (XAI)-driven methodology to enhance the understanding of cough sound analysis for respiratory disease management. We employ occlusion maps to highlight relevant spectral regions in cough spectrograms processed by a Convolutional Neural Network (CNN). Subsequently, spectral analysis of spectrograms weighted by these occlusion maps reveals significant differences between disease groups, particularly in patients with COPD, where cough patterns appear more variable in the identified spectral regions of interest. This contrasts with the lack of significant differences observed when analyzing raw spectrograms. The proposed approach extracts and analyzes several spectral features, demonstrating the potential of XAI techniques to uncover disease-specific acoustic signatures and improve the diagnostic capabilities of cough sound analysis by providing more interpretable results.

XAI-Driven Spectral Analysis of Cough Sounds for Respiratory Disease Characterization

TL;DR

The paper addresses the challenge of disease differentiation from cough sounds by introducing an eXplainable AI (XAI) workflow that uses occlusion maps on CNN-derived cough spectrograms to weight spectral information. By extracting spectral features from these weighted spectrograms, the study demonstrates significant differences between disease groups, most notably COPD, which shows greater spectral variability than other pathologies. In contrast, analyses on unweighted, raw spectrograms fail to reveal such distinctions, highlighting the value of XAI-guided feature extraction. The approach improves interpretability and has potential to enhance non-invasive, cough-based diagnostics for respiratory diseases.

Abstract

This paper proposes an eXplainable Artificial Intelligence (XAI)-driven methodology to enhance the understanding of cough sound analysis for respiratory disease management. We employ occlusion maps to highlight relevant spectral regions in cough spectrograms processed by a Convolutional Neural Network (CNN). Subsequently, spectral analysis of spectrograms weighted by these occlusion maps reveals significant differences between disease groups, particularly in patients with COPD, where cough patterns appear more variable in the identified spectral regions of interest. This contrasts with the lack of significant differences observed when analyzing raw spectrograms. The proposed approach extracts and analyzes several spectral features, demonstrating the potential of XAI techniques to uncover disease-specific acoustic signatures and improve the diagnostic capabilities of cough sound analysis by providing more interpretable results.

Paper Structure

This paper contains 16 sections, 9 equations, 10 figures, 4 tables.

Table of Contents

  1. Introduction
  2. State of The Art
  3. ML-based Cough detection from hand-crafted features
  4. Cough detection using DL
  5. Cough analysis for diagnosis support
  6. Addressing Research Gaps in Cough Analysis: Using XAI for Disease Differentiation
  7. Materials and methods
  8. Materials
  9. Methods
  10. Audio signal preprocessing
  11. Cough Window Identification
  12. XAI-driven cough analysis
  13. Spectral features obtained from the Weighted Spectrograms
  14. Testing for group differences
  15. Results and Discussion
  16. ...and 1 more sections

Figures (10)

  • Figure 1: CNN architecture for cough detection.
  • Figure 2: Median Weighted Spectrogram for the pathology sets
  • Figure 3: Overview of the classification and XAI-driven cough analysis.
  • Figure 4: Boxplots obtained for Relative AC Power ($AC$) at the most discriminative threshold.
  • Figure 5: Boxplots obtained for Spectral Bandwidth ($SpBw$) at the most discriminative threshold.
  • ...and 5 more figures