MING PULMONARY SOUND DIAGNOSTICS: UTILIZING PRETRAINED CONVOLUTIONAL NEURAL NETWORKS FOR DETECTION

Abstract

Lung sound signals, a vital physiological indicator in clinical auscultation, contain a wealth of information crucial for diagnosing and monitoring human health. In the realm of clinical auscultation, doctors have traditionally relied on their subjective judgment to categorize patients' lung sounds, aiding in the diagnosis of lung diseases. However, this subjective approach can lead to missed diagnoses and misdiagnoses, posing significant risks to patients. While X-ray methods like chest X-rays and lung function tests are widely employed, they involve harmful radiation. In contrast, lung sound auscultation poses no additional risk to the human body. With advancements in computer science, a promising avenue is emerging: the use of machine learning techniques to recognize and classify lung sound signals, thereby assisting healthcare professionals in diagnosing pulmonary diseases without additional harm.