This study aims to use the three-dimensional computed tomography as a diagnostic tool for the preoperative evaluation of the pathology analysis of the middle and inner ear. Our goal is to setup an automatic computer-based system which is able to recognize the bone contours such as the temporal bone on the CT images, so that we can have a 3D view of the temporal bone before the operation is performed. The commercial products currently used in Radiology Department are unable to perform the image segmentation according to physician’s demand. Such images usually need to be manually corrected after primitive segmentation, and therefore result in time-consuming manual work. Moreover, the image result cannot be completely redone because of such manual works (intra-observer variation). Different segmentation results, depending on the physician’s judgment, have inter-observer variations. This study uses the image processing technique with embedded optimized automatic image segmentation of anatomic knowledge and 3D reconstruction process.
