A Brief Overview of the Respiratory Sound Based Detection of Chronic Obstructive Pulmonary Disease(COPD)

Lung sound reflects the lung morphology. Changes in behavior of respiratory sound could be the direct indicator of changes in the lung morphology due to pulmonary diseases. Computerized lung sound analysis is one of the cost effective as well as the non-invasive tool to analyze respiratory diseases. Chronic Obstructive Pulmonary Disease (COPD) is one of the chronic lung diseases whose correct diagnosis is very difficult as well as whose health cost is very high. Researchers have proposed several interesting methodologies in diagnosing such disease based on LS. However, it is essential to identify the available lung sound (LS) based diagnostic system which could be able to effectively detect COPD in a better way. This study is all about to identify the exact LS based diagnostic procedure which could be a considerable and effective one for diagnosing COPD. In comparison to all of the existing LS based methodologies, one study produced the highest classification accuracy of 99.7%. In that the study the LS recordings were effectively denoised using EMD and DWT, followed by feature extraction in various domains. Next to this, feature selection was conducted using the back elimination method to identify the best features. Those selected features were then given input to quadratic discriminant classifier to diagnose COPD. This LS based diagnostic procedure outperformed with classification accuracy of 99.7% It is concluded that this study has successfully identified the most effective automated LS based procedure to correctly detect COPD. This methodology can even be recommended clinically as to support physicians for diagnosing COPD. However, existing studies have not included the LS data of COPD patients having other comorbidities. The future endeavor of this study is to implement such a LS based diagnostic system which could detect COPD even in COPD patients with other comorbidities.