Development of a Spectral Feature Extraction using Enhanced MFCC for Respiratory Sound Analysis

Chronic illnesses such as respiratory diseases are among the most persistent health threats in our society nowadays. Fortunately, the emergence of state-of-the-art technologies like Internet of Things (IoT), Machine Learning, and Artificial Intelligence (AI) are available to make monitoring and pre-diagnosis of human health conditions fast and convenient. Nowadays, health services that are accurate, accessible, and convenient are amongst the in-demand in modern medical applications. In this study, an efficient design for a lung sound classifier is explored that utilizes enhanced-Mel frequency cepstral coefficients (eMFCC). Spectral feature extraction based on MFCC is implemented and optimized using MATLAB. MFCC parameters such as frame duration, frameshift, number of filterbank channels, number of cepstral coefficients, and the frequency range are included in this study. The enhanced MFCC feature vectors were extracted using a histogram and were subjected to different machine learning algorithms such as Support Vector Machine (SVM) and K-Nearest Neighbors (KNN). Results show the evaluation of the enhanced MFCC based on sensitivity, specificity, and overall accuracy is higher than the conventional MFCC.