B79-19 Lung Sound Analysis as a Noninvasive Tool for Assessing Disease Severity in Idiopathic Pulmonary Fibrosis

Rationale

Idiopathic pulmonary fibrosis (IPF) is a heterogeneous fibrotic lung disease in which disease severity and progression are most commonly assessed using computed tomography (CT). Although CT plays a central role in diagnosis and staging, its repeated use for longitudinal monitoring is limited by radiation exposure, cost, and practical constraints. Lung crackles, a characteristic auscultatory feature of IPF, may provide complementary, noninvasive information reflecting underlying structural and functional alterations in the fibrotic lung. Analysis of lung sound recordings represents a supplementary approach for monitoring disease progression. Unlike CT imaging, this method avoids radiation exposure and can provide quantitative information on crackle characteristics, including their timing, duration, and energy distribution over time.

Methods

Lung sound recordings were obtained from seven patients with CT-confirmed IPF. Audio signals were preprocessed using band-pass filtering to reduce background noise and cardiac interference. Crackle events were automatically detected using a time-domain approach targeting short-duration acoustic transients (Figure 1A). Temporal and spectral features extracted included crackle rate, coarse crackle proportion (ratio of coarse to total crackles), median crackle duration, and median spectral centroid frequency. CT findings were used solely as a reference framework to aid interpretation of acoustic features (Figure 1B).

Results

Fifteen-second lung sound recordings were obtained from all seven patients. As Figure1-C indicates, although crackle rate generally increased with disease severity, substantial interpatient variability was observed. Notably, Patient 6 (severe IPF) demonstrated a lower crackle rate than Patient 5 (moderate IPF), despite exhibiting longer crackle duration and higher spectral centroid frequency. This pattern suggests that advanced fibrotic remodeling may reduce crackle generation while increasing tissue stiffness. Overall, these findings indicate that a combination of acoustic features, rather than crackle rate alone, is more informative for characterizing disease stage.

Conclusions

Lung sound analysis provides useful, noninvasive information related to disease stage in patients with idiopathic pulmonary fibrosis. The diversity of acoustic parameters derived from lung sounds may assist clinicians in therapeutic decision-making. In addition, disease progression can be monitored by longitudinal comparison of these parameters, potentially reducing reliance on repeated, costly CT imaging. With sufficiently large datasets, machine learning approaches could be applied to lung sound features to support disease staging and progression assessment without the need for routine CT scans.

This abstract is funded by: None