Chapter 9 - Computational lung sound classification: a review

Abstract

Computational lung sound classification (LSC) has attracted attention over many years. The number of studies on adventious lung sound detection as well as respiratory disease classification has increased dramatically. This chapter summarizes existing approaches of conventional machine learning and deep learning-based LSC systems. We provide a structural review of LSC systems including topics from data processing such as audio signal processing, feature extraction, and data augmentation to data modeling such as neural network architectures and learning paradigms. In addition, we shortly discuss current advances and open challenges for possible future developments toward the real-life application of LSC systems.