Segmentation techniques for the extraction of cranial aneurysms

Many people in the United States suffer from sub-arachnoidal hemorrhages caused by the formation of cranial aneurysms. Many die even before they reach the hospital. Of the people who survive, many suffer from stroke from the emboli (blood clot) formed near the aneurysm. In this thesis, semi-automatic segmentation techniques are presented to extract cranial aneurysms from CTA data sets. A simple region-growing technique is initially used for segmenting berry type aneurysms. A sub-volume is used to extract only the aneurysm and the connecting vessels. This technique is used in combination with the morphological methods to eliminate the effects of the partial volume effect. A bone removal technique is also presented to improve the results of the combined region growing and morphological technique. The results of the technique are used to create a 3D model of the aneurysm. The models were used in performing computational fluid dynamics calculations. The technique presented in this thesis also facilitates viewing and analysis of the aneurysm with less manual interaction and the visualization and quantitative results can be generated in a few minutes.