Advancing Auscultation Education: Signals Visualization as a Novel Tool for Enhancing Pathological Respiratory Sounds Detection

Introduction: Auscultation remains a pivotal diagnostic modality for various respiratory pathologies. To augment its clinical relevance, the continuous expansion of our understanding of pulmonary acoustics, coupled with the advancement of auscultation recording and analysis methodologies, is imperative. Material and methods: We investigated how the bimodal presentation of auscultatory signals (sound and visual cue perception) influences the subjective efficacy of pathological respiratory sound detection, which is a critical step in the development of a new auscultation tool. Recordings of pediatric breath sounds were presented in three different forms - audio only, visual representation only (spectrogram) or audiovisual (both together). The F1-score, sensitivity and specificity parameters were calculated and compared to the gold standard (GS). Subsequent to the detection experiment, participants completed a survey to subjectively assess the usability of spectrograms in the procedure. Results: Over 60% of all responders ranked the spectrogram as important (40.8%) or very important (21.1%). Moreover, 11.3% of all participants found this new form of presentation of auscultation results to be more useful than the evaluation of sound only. The addition of visual information did not statistically significantly change the evaluation of the sounds themselves, an observable trend implies that enhancing audio recordings with visualizations can enhance detection performance. This is evident in the 4 p.p. and 2 p.p. sensitivity increments for physicians and students, respectively, even without specialized visual training. Conclusions: Our research findings indicate that the integration of spectrograms with conventional auditory assessment, albeit based on observed trends and survey responses, presents a promising avenue for improving the precision and quality of medical education, as well as enhancing diagnosis and monitoring processes.