SU‐E‐J‐23: Automatic Detection of Spherical Gold Fiducials in KV X‐Ray Images Having Scattered KV and MV X‐Ray

Abstract

Purpose: In tumor tracking irradiation using Vero4DRT having two sets of a kV x‐ray tube and a flat panel detector (FPD), positions of a tumor are estimated from positions of spherical gold fiducials implanted closely to the tumor. The purpose of this study was to propose a new fiducial detection algorithm in kV x‐ray images having scattered kV and MV x‐ray. Methods: According to the preliminary experiment using a fiducial of 1.5‐mm diameter placed on a water‐equivalent phantom and a single set of a kV x‐ray tube and an FPD, we have observed that the logarithm of a ratio of the background to the fiducial intensity decreased in proportion to the x‐ray path length and that the background intensity was proportion to the inverse of x‐ray path length. By combining the above two regression equations and adding a scattered radiation term by the MV and/or the other kV x‐ray sources, the fiducial intensity could be described as a function of the background intensity and the scattered radiation amount without the x‐ray path length. Here, the scattered radiation amount was given as the minimum value allowing all fiducials to be detected. Finally, the fiducials were specified using the fiducial intensity estimated from the above function and some other features such as the lower limit intensity for non‐fiducials. The detection accuracy was measured using 3,824 fiducials in 900 pairs (for both sets of x‐ray tubes and FPDs) of kV x‐ray image with a resolution of 0.21 mm/pixel at the isocenter. Results: The algorithm detected 99.50% of the fiducials. The positional errors were 0.16, 0.16, and 0.21 mm in left‐right, cranio‐caudal, and anterior‐posterior direction, respectively. Conclusion: We have demonstrated that our proposed fiducial detection algorithm has a high and robust accuracy for the kV x‐ray images with scattered radiation. This research is granted by the Japan Society for the Promotion of Science through the Funding Program for World‐Learning Innovative R&D on Science and Technology, initiated by the Council for Science and Technology Policy. This research was in part sponsored by Mitsubishi Heavy Industries, Ltd., Japan. Akira Sawada, Takashi Mizowaki, Masaki Kokubo, and Masahiro Hiraoka have consultancy agreement with Mitsubishi Heavy Industries, Ltd., Japan. Kunio Takahashi is an employee of Mitsubishi Heavy Industries, Ltd., Japan.