Abstract:
An algorithm for the reconstruction of the distribution of sound sources within the thorax was developed. For any point within the thorax, the algorithm tests the hypothesis that it contains the only relevant source to explain a given set of simultaneous measurements on the chest wall. A hypothetical source signal is calculated by a least-squares estimation procedure for all points 1 cm apart on a equally spaced grid overlying the thoracic volume. The explained fraction of the total signal variance of all microphones is then displayed as a function of space in a 3-dimensional grey-scale coded image. The algorithm was applied to simultaneous lung sound recordings with eight microphones from two healthy male subjects. Preliminary findings suggest that inspiratory sounds are produced more peripherally in the thorax, while expiratory sounds originate more centrally. Some of the images depict features which correlate with anatomical structures such as the large airways, the heart, or during the process of swallowing, the esophagus. A mechanical lung model is being developed to investigate the significance and reproducibility of these images features. [Work supported by Swiss National Research Foundation, Roche Research Foundation, and NSF.]