eduzhai > Applied Sciences > Engineering >

A Lightweight CNN Model for Detecting Respiratory Diseases from Lung Auscultation Sounds using EMD-CWT-based Hybrid Scalogram

  • king
  • (0) Download
  • 20210507
  • Save

... pages left unread,continue reading

Document pages: 9 pages

Abstract: Listening to lung sounds through auscultation is vital in examining therespiratory system for abnormalities. Automated analysis of lung auscultationsounds can be beneficial to the health systems in low-resource settings wherethere is a lack of skilled physicians. In this work, we propose a lightweightconvolutional neural network (CNN) architecture to classify respiratorydiseases using hybrid scalogram-based features of lung sounds. The hybridscalogram features utilize the empirical mode decomposition (EMD) andcontinuous wavelet transform (CWT). The proposed scheme s performance isstudied using a patient independent train-validation set from the publiclyavailable ICBHI 2017 lung sound dataset. Employing the proposed framework,weighted accuracy scores of 99.20 for ternary chronic classification and99.05 for six-class pathological classification are achieved, which outperformwell-known and much larger VGG16 in terms of accuracy by 0.52 and 1.77 respectively. The proposed CNN model also outperforms other contemporarylightweight models while being computationally comparable.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×