Research On Lung Sound Recognition Based On Hilbert-Huang Transform

Lung sound signal is a physiological acoustic signals generated outside the human respiratory system and in the ventilation process. It contains a wealth of physiological and pathological information and has great value in lung sounds study and clinic medicine. Although traditional manual auscultatory method for medical diagnosis of lung disease is widely popular because of its convenience non-invasive properties, there are also subjectivity by human judgment and the limitation in equipments and human hearing. Therefore, the electronic and automatic recognition of lung sound will be the inevitable development trend in this field.In recent years, research in lung sound developed rapidly with the progress in modern signal analysis techniques, the hardware facilities, and the knowledge of lung sound in different diseases. However, there are still some great limitations and shortcomings: 1. Results after the recognition and classification based on machine or computer only show whether the lung sound is normal or abnormal, or divided into wheeze or coarse. The results lack of specific pathological information to combine with actual treatment, such as the position in respiratory phases where the abnormal sound exists on. 2. Unlike ECG and EEG, lung sound is still dependent on the experience of the physician’s artificial auditory judgment for further diagnosis rather than visual diagnosis. In this paper a method to segment the respiratory phase based on Hilbert-Huang transform has been proposed, and a software system contains the respiratory phase segmentation of lung sound has been built.Three aspects are included in this paper:(1) The knowledge of lung sound are systematically introduced and the features of typical abnormal lung sound are explained. Then the current method for lung sounds recognition techniques are discussed.(2) The basic concept and theory of Hilbert-Huang transform method are introduced with a vast amount of literature reading and study. Then a method based on Hilbert-Huang transform in lung sound respiratory phase segmentation areproposed(3) A lung sound recognition software system includes respiratory phase segmentation are built through MATLAB. The system can segment the respiratory phase semi-automatically and build the features in lung sounds recognition. Besides,it also has good usability in visual aids diagnosis.