Spectral Analysis of Lungs sounds for Classification of Asthma and Pneumonia Wheezing
Syed Zohaib Hassan Naqvi, Misha Arooj, Sumair Aziz, Muhammad Umar Khan, Mohammad Ahmad Choudhary, Muhammad Nafees. ul. Hassan
Abstract
World Health Organization Statistics declares the pulmonic illness as the class of deadly illness. Wheezing is a key indicator for the diagnosis of pulmonic illnesses like Asthma and pneumonia. In this research article, the identification of wheeze sound in asthma and pneumonia subjects is done from breathing sound. The analysis is performed through signal processing and machine learning practices. Overall, data is acquired from 300 subjects. It includes 100 Asthma, 100 Pneumonia, and 100 Normal subjects This research work proposes a complete design for accurate classification of wheezing signals. It includes pre-processing by normalization, denoising by filtration, segmentation to remove the non-breathing and silent parts, feature extraction from the spectral domain, and classification by support vector machine (SVM) using Matlab 2019b. The system evidenced an accuracy greater than 96%. Further investigation can be done by analyzing the wheezing sound originates in other pulmonic diseases and exploring its role to identify the pulmonary illness.