Scaled attenuation fields: Improved realtime generation method for digitally reconstructed radiographs

Abstract

Image guidance technology has made it possible for a system to track the location of a patient’s bones or lesions in real time while the patient undergoes medical treatment, with a specific part of the patient’s body serving as the reference. To this end, intraoperative 2D X-ray images and preoperative 3D CT images are continuously registered at specific intervals. The implementation of such a system requires a quick and efficient generation algorithm for digitally reconstructed radiograph (DRR). Since the generation of a DRR from CT by ray casting has a time complexity of O(n3), it cannot be generated at a speed sufficient for real-time tracking of the location of the patient during an operation. As a solution to this problem, this study proposes scaled attenuation fields (SAFs). In this study, the time complexity of the DRR generation process is reduced to O(n2), thus allowing a speed sufficiently fast for real-time image registration. In addition, by scaling CT data while generating the attenuation fields, bones, which serve as the reference for image registration, can be more clearly distinguished from fat or muscles, thus improving the accuracy of the system.