Respiratory sound analysis using AI/ machine learning algorithms

• Main Author: Zhang, Jingyi
• Other Authors: Ser Wee
• Format: Thesis-Master by Coursework
• Language: English
• Published: Nanyang Technological University 2020
• Subjects: Engineering::Electrical and electronic engineering
• Online Access: https://hdl.handle.net/10356/141445
• Institution: Nanyang Technological University

In recent years, AI and machine learning techniques have been applied extensively in many biomedical and healthcare applications. Cardiopulmonary diseases are prevailing diseases and the leading causes of death in many parts of the world. Respiratory sounds are generated by the air flow going in and out of the airways. A dysfunction in the respiratory or cardiovascular system will generate sounds of different characteristics from that of a normal one. As such, pulmonary auscultation has been an extremely valuable tool in the diagnosis of respiratory and cardiovascular conditions. Recently, automated respiratory sound analysis using AI or machine learning algorithms have received much attention from both the academia and the industry. The primary objective of this project is to develop signal classification algorithms that are able to distinguish between normal and abnormal respiratory sounds accurately. For this propose of the project, the respiratory signal will be re-processed firstly. Then prepare the label and extract the feature by calculating the MFCC feature. The primary classification task in the project is to classify the normal and abnormal respiratory sounds. In classification process, SVM model and CNN model are used to do the classification and compare the performance of them. As the result, the classification accuracy on SVM and CNN can achieve 0.9324 and 0.8889 respectively. This shows that the algorithm presented in this report has high accuracy and good prospects.