Electronic auscultation based on wavelet transformation in clinical use

Electronic auscultation is an important, non-invasive method for the detection of lung diseases. For computerized lung sound analysis the wavelet transformation (WT) is a good tool to detect and analyze discontinuous and continuous pathological lung sounds. We developed a lung sound analysis system using a WT based combined method of filtering and spectral analysis to extract parameters from lung sound recordings that could detect typical pneumonia lung sounds (crackles, bronchial breathing). The lung sounds of 32 patients with pneumonia were recorded simultaneously with two air-coupled microphones at standardized airflow. We computed parameters to detect crackles and bronchial breathing sound patterns. Our parameters are able to detect typical pneumonia lung sounds at an early stage of the disease in all 32 patients. Our system found automatically a bronchial breathing sound pattern in 16 patients and crackles in 20 patients as confirmed by normal auscultation findings. As the procedure of electronic auscultation is neither invasive nor particularly expensive, it would be suitable as an examination method for critical care patients who have a high risk of getting hospital acquired pneumonia.