DETERMINATION OF LUNG SOUND AS NORMAL OR ABNORMAL, USING A STATISTICAL TECHNIQUE

  • Isuri Liyanage Department of Electronic, Electrical and Telecommunication Engineering, General Sir John Kotelawala Defence University
  • P.K.G.Y Siriwardhana Department of Electronic, Electrical and Telecommunication Engineering, General Sir John Kotelawala Defence University.
  • W.H.A.U Abeyrathne Department of Electronic, Electrical and Telecommunication Engineering, General Sir John Kotelawala Defence University
Keywords: Adaptive Noise Cancellation, Least Mean Square Algorithm, Lung Sound Analysis, Mahanaobis Distance

Abstract

In this study the authors investigate a possibility of objectively differentiating a lung sound as normal or abnormal using a statistical technique. For the study, breath sounds were recorded from 30 nonsmoking, healthy subjects and 7 subjects with respiratory disorders whose external physical symptoms were not shown, using an electronic stethoscope. A 4th order Butterworth bandpass filter removed environment sounds and an Adaptive filter using Least Mean Square algorithm cancelled other body sounds from the recorded sound to obtain only the lung sound. After amplifying a lung sound signal up to the initial recorded amplitude, signal was compared with a standard normal and a standard abnormal lung sound. The comparison was done by calculating the Mahalanobis Distance mean values. The Mahalanobis distance mean values obtained from subjects with respiratory disorders showed considerable deviations from the specific range of values obtained by subjects with normal lung sounds concluding this method is capable of distinguishing between normal and abnormal lung sounds andĀ  couldĀ  developed to noninvasively determine the progress of patients with respiratory disorders.

Published
2017-09-13
How to Cite
Liyanage, I., Siriwardhana, P., & Abeyrathne, W. (2017). DETERMINATION OF LUNG SOUND AS NORMAL OR ABNORMAL, USING A STATISTICAL TECHNIQUE. Proceedings of International Conference on BioScience and Biotechnology, 2(1), 14-23. https://doi.org/10.17501/biotech.2017.2102