Adaptive Noise Cancellation For Electronic Stethoscopes
Translated title
Adaptiv støjreduktion til elektroniske stetoskoper
Author
Term
4. semester
Education
Publication year
2025
Submitted on
2025-05-28
Pages
150
Abstract
This thesis investigates adaptive noise cancellation for improving heart and lung sound recordings in noisy environments, with a focus on resource-limited settings like rural India. Environmental and equipment noise often degrade auscultation quality, affecting diagnosis. To address this, adaptive filters (LMS, NLMS, RLS) and Independent Component Analysis (FastICA) were evaluated. All methods met the target of improving SNR by at least 20 dB in synthetically mixed signals, confirming their theoretical effectiveness. RLS stood out among the filters, achieving up to 34 dB improvement and performing reliably across noise types using fixed hyperparameter. While FastICA showed strong results in controlled conditions, it was unreliable in real-world recordings due to timing and environmental sensitivity. In contrast, RLS with fixed hyperparameter proved practical and robust for real-life use. These results highlight RLS-based adaptive filtering as a simple, effective solution for enhancing auscultation in noisy conditions, supporting broader deployment in frontline health settings.
Documents