MR Image Reconstruction for Positioning Verification with a Virtual Simulation System for Radiation Therapy

Abstract

This paper proposes a systematic method for 3D reconstruction of the MR images of a head to create a virtual radiation target for positioning verification in the treatment planning of radiotherapy. According to different features in the scalp and the brain, two different methods were applied to segment the entities. The scalp was detected by auto-thresholding and the brain was segmented by region-growing. The detected contours were then reconstructed as the 3D polygon meshes and transformed into VRML format to be presented in the virtual simulation environment. This scheme is further implemented as a module embedded in a virtual simulation system of 3D Conformal Radiotherapy (3D-CRT). System verification was achieved by using a full scale AP model of a human head as a target volume for positioning verification. We compared the positioning of the AP model in the real and virtual systems and found that the measurements were very close (difference < 0.3 cm). This state of the art integrates the techniques of image processing, computer graphics and virtual reality. With the assistance of the system, the position verification process may become more efficient and accurate than the conventional positioning toolkit only based on 2D image registration.