Lung Sounds Classification Based on Time Domain Features

Abstract

Signal complexity in lung sounds is assumed to be able to differentiate and classify characteristic lung sound between normal and abnormal in most cases. Previous research has employed a variety of modification approaches to obtain lung sound features. In contrast to earlier research, time-domain features were used to extract features in lung sound classification. Electromyogram (EMG) signal analysis frequently employs this time-domain characteristic. Time-domain features are MAV, SSI, Var, RMS, LOG, WL, AAC, DASDV, and AFB. The benefit of this method is that it allows for direct feature extraction without the requirement for transformation. Several classifiers were used to examine five different types of lung sound data. The highest accuracy was 93.9 percent, obtained Using the decision tree with 9 types of time-domain features. The proposed method could extract features from lung sounds as an alternative.