Recognizing Thoracic Breathing by Ensemble Empirical Mode Decomposition

Abstract

Recognizing breathing pattern is important in many fields of medicine. Ensemble empirical mode decomposition (an adaptive algorithm) was used to investigate breathing pattern, including thoracic breathing (TB) and abdominal breathing (AB). This study recognizes TB and AB by correlation coefficient and power proportion. Results indicate that the recognition accuracy of TB by correlation coefficient and power proportion are 85.2% and 93.3% respectively, and that of AB by correlation coefficient and power proportion are 54.3% and 56.2% respectively. The TB can be well defined and recognized in complex time variation. These results can be used as references to develop the real time breathing evaluation system in the future.