Detection of Airway Obstruction from Frequency Distribution Feature of Lung Sounds with Small Power of Abnormal Sounds
We propose a new method to detect airway obstruction from a lung sound record with power of which abnormal sounds is much smaller than power of normal sound s. One of traditional methods to detect airway obstruction is FEV1% (forced expiratory volume 1 sec percentage) using a spirometry. But it bothers a patient too much. Some methods were proposed recently to detect abnormal sounds because an airway obstruction sometimes makes abnormal sounds such as wheeze or rhonchi or else. But it is not available for cases with small power of abnormal sounds. The correlation coefficient between our proposed value and FEV1% was -.592. And the AUC value of the proposed method with 70% threshold of FEV1% was 0.833. The proposed method could detect airway obstruction with sensitivity=0.8 and specificity = 0.78 FEV1%.
KeywordsBronchial asthma Airway constriction FEV1% Fourier transform Wheeze Diagnosis
Unable to display preview. Download preview PDF.
- 1.Ortiz, G.: Asthma Diagnosis and Management: A Review of the Updated National Asthma Education and Prevention Program Treatment Guidelines. The Internet Journal of Academic Physician Assistants 6(2) (2008)Google Scholar
- 10.Bahoura, M., Hubin, M.: Automatic wheeze detection using wavelet packets. In: Conference on Medical and Biological Engineering and Computing VIII Mediterranean, Limassol, Cyprus, pp. 14–17, June 1998Google Scholar
- 11.Bahoura, M., Lu, X.: Separation of crackles from respiratory sounds using wavelet packet transform. In: The 31st International Conference on Acoustics, Speech, and Signal Processing (ICASSP-06), Toulouse, France, vol. II, pp. 1076–1079, May 2006Google Scholar