Multiscale Morphological Operators and Geodesic Distance applied to Computed Axial Tomography Segmentation

Lately, several diagnostic and therapeutic procedures for mediastinal tumors and masses are done by videothoraxcoscopic techniques. Consequently, Computed Axial Tomography (C.A.T.) become a fundamental tool in tumors and masses diagnostic, especially to locate and measure them with accuracy and precision. This paper presents a segmentation method in C.A.T. images through techniques based in Mathematical Morphology. A semi-automatic method for structure segmentation based on Multiscale Morphological Operators is introduced. In addition to filtering and enhancing, this method capture the interior of a closed simple curve by the geodesic distance. These curves represents the external boundary of the different mediastinal structures. This method allows to calculate exactly the external boundary of the segmented structure and also the delimitated area.