[Automatic boundary detection and pathological diagnosis of the maxillary sinus]

A method for automatic boundary detection of the maxillary sinus from a cranial X-ray radiograph has been developed. An algorithm using the local segmental properties of the X-ray image to transform a continuous tone image into a bi-level image is introduced. In addition, the global shape information such as the area, the perimeter, and the complexity, are considered for the boundary detection. The boundary of the maxillary sinus detected by this method coincided precisely with the area plotted-by experienced otolaryngologists. Furthermore, the cumulative distribution function (CDF) calculated from the density histogram of the maxillary sinus has been employed to perform the tonal evaluation of the intra-maxillary pathological changes. The distance between the CDFs successfully demonstrated the difference in pathological stages of the chronic sinusitis.