부산대학교 도서관

In this paper, a computer aided system is developed for analyzing and visualizing CT images with nasopharyngeal carcinoma. In the system, firstly, an effective registration method that combines global rigid registration algorithm and local multi-resolution symmetric Demons registration algorithm is applied to match the plain CT image and the contrast-enhanced CT image of the same patient. Secondly, a total probability map is generated to roughly locate the tumor and determine an initial seed in the tumor. Here, this map is derived from CT intensities of the plain CT image and the contrast-enhanced CT image of the same patient and from location of the nasopharynx. Thirdly, a region growing method with multi feature criteria is used to segment the tumor from the volume data. Finally, the tumor and head skin are reconstructed and visualized so that the size and location of tumor are displayed. The experimental results show that this system can be used as a medical auxiliary diagnostic tool for radiologists to observe the size and location of the tumor.