CT MRI Registration Based on Images' Geometrical Features
Zhao Fuqiang, Yu Hongyu, Ma Binrong, Li Kuncheng (Capital University of Medical Science(100054)
In this paper is described an automated approach to register CT and MRI brain images based on their geometrical features. Local second order differential operators in scale space are applied to each type image date, so as to produce feature images, depicting ridges or vallegs. Then correlation calculation is carried out for feature images under the condition of rigidity, so that the coeffiuent of correlation is maximum. The resulted CT with MRI feature images are then obtained. This method does not need human interaction and interactive operation is also not needed during post processing. Results obtained using this method are presented. It is proved that this method is practically useful.