A Systematic Review of Transfer Learning Methods for Identifying Lung Disease Sounds

Abstract

Lung disorders are now the leading cause of death throughout the globe. Despite this, most occurrences of lung illness are only identified at a late stage, when treatment options may be more limited. Technological advances are crucial to today’s healthcare delivery system. This state-of-the-art medical research focuses on the value of analyzing lung sounds for the purpose of identifying lung diseases. The capacity to learn new material and use it in novel situations is crucial for patients to make their way through the healthcare system. Several Transfer learning techniques, like ALEXNET, VGGNET, and RESNET, are presented in this paper for classifying lung sounds. In addition to these methods, we will classify lung sound waves using a Transfer learning model that combines a Modified RESNET and a Mel spectrogram. Excellent performance in categorizing lung sounds by these transfer learning models suggests they may one day be employed in the diagnosis of respiratory disorders. In this evaluation, we will look at several Transfer Learning Techniques and talk about their pros and cons. And not even the worst part. To recognize four kinds of breathing noises. In addition, please provide suggestions about how the identification of lung sounds might” be improved.