Lung Disease Diagnosis using Sound Processing and Machine Learning

Abstract

Automated and non-intrusive lung sound monitoring can significantly enhance the management of respiratory diseases such as chronic obstructive pulmonary disease (COPD), pneumonia, asthma, and Covid-19. The auscultated lung sound, typically monitored using a stethoscope, contains very rich diagnostic information. However, its contamination with other sound sources such as noise and vocal, heart, and movement-related sounds, does not allow an accurate automation of lung diagnosis from a single channel auscultation. Hence, the overarching aim of this research is to develop machine learning algorithms which will enable real-time sound monitoring by extracting the lung sound from its noisy mixtures, and subsequently extracting features for its classification. The machine learning system to be developed in this project will predict any upcoming alarming situations such as asthma, heart failure, pneumonia, bronchitis, pleural effusion, ling fibrosis, and COPD.

Date of Award	7 May 2024
Original language	American English
Supervisor	PANAGIOTIS Liatsis (Supervisor)