Detecting changes in respiratory patterns in high frequency chest compression therapy by single-channel blind source separation

High frequency chest compression (HFCC) is used as a method to remove the mucus in the airway for cystic fibrosis (CF) patients. As the characteristics of the tracheal sound reflect the conditions of airways, in this paper, we propose a novel method to evaluate the respiratory patterns in HFCC therapy by using single channel tracheal sounds only. The difficulty of analyzing tracheal sounds lies in that it has a wider frequency band than the air flow at the mouth, and is always corrupted by other biomedical signals and noises. During HFCC therapy, the tracheal sound is also affected by the HFCC machine noise. For this reason, it is difficult to extract respiratory patterns and other related features by traditional filtering techniques. In this paper, we demonstrate use of single-channel independent component analysis to extract respiratory patterns from the tracheal sounds before, during and after HFCC therapy, and use basis features in the tracheal sound to detect the change in respiratory patterns.