Speaker: Miguel Ángel Martínez Pay.
Abstract: Based on https://ieeexplore.ieee.org/document/10051156. Wheezes, a respiratory anomaly in patients with respiratory conditions, are significant for clinical assessment, particularly in gauging bronchial obstruction. While conventional auscultation is the norm for wheeze analysis, recent years emphasize the need for remote monitoring. Automatic analysis of respiratory sounds is essential for reliable remote auscultation. This study proposes a method for wheeze segmentation, beginning with audio decomposition into intrinsic mode frequencies using empirical mode decomposition. Harmonic-percussive source separation and subsequent processing yield harmonic-enhanced spectrograms and masks. Empirically derived rules are then used to identify wheeze candidates. Finally, these candidates from various audio tracks are merged and median filtered. The evaluation demonstrates the method’s superiority over baselines on the challenging ICBHI 2017 Respiratory Sound Database. The conclusion underscores the unresolved nature of wheeze segmentation for real-life applications, suggesting potential in adapting systems to demographic traits for algorithm personalization, rendering automatic wheeze segmentation clinically viable.