Effect of Window Size for Detection of Abnormalities in Respiratory Sounds

TL;DR

This study investigates how different window sizes affect the accuracy of detecting respiratory abnormalities, finding that 2 to 10 seconds windows yield optimal performance with over 92% accuracy.

Contribution

It analyzes the impact of window size on respiratory sound diagnosis and identifies the most efficient recording duration for accurate detection.

Findings

Optimal window size is between 2 and 10 seconds.

Performance reaches an average of 92.06%.

Shorter or longer windows reduce accuracy.

Abstract

The recording of respiratory sounds was of significant benefit in the diagnosis of abnormalities in respiratory sounds. The duration of the sounds used in the diagnosis affects the speed of the diagnosis. In this study, the effect of window size on diagnosis of abnormalities in respiratory sounds and the most efficient recording time for diagnosis were analyzed. First, window size was applied to each sound in the data set consisting of normal and abnormal breathing sounds, 0.5 second and from 1 to 20 seconds Increased by 1 second. Then, the data applied to window size was inferred feature extraction with Mel Frequency Cepstral Coefficient (MFCC) and the performance of each window was calculated using the leave one-out classifier and the k-nearest neighbor (KNN) algorithm. As a result, it was determined that the data was significant with an average performance of 92.06% in the records between 2 and 10 seconds.