| Summary: Lung sounds are changeable by some respiratory defections of human. In order to examine these sounds, a system was developed which called auscultation. In this system, used to hear lung sounds as a non-invasive device named stethoscope. The number of studies has increased in this area while growing medical electronic. Analogue lung sounds taken by an electronic stethoscope and saved a computer via digital voice recorder and practised to get the best way of classification of these sounds while MFKK coefficients are the main parameter of some feature extraction paths. k-NN and SVM used as a classifier. According to this studys results, the combination with the highest classification accuracy performance is the set of attributes that calculated by taking the average of MFCC coefficients and calculated lung sounds. After that, the classification performance comes as standard deviation and entropy. |
