Naoki Miyamoto1, Kenichiro Maeda2, Daisuke Abo3, Ryo Morita3, Seishin Takao4, Taeko Matsuura1, Norio Katoh5, Kikuo Umegaki1, Shinichi Shimizu6, Hiroki Shirato5. 1. Division of Quantum Science and Engineering, Faculty of Engineering, Hokkaido University, North 13, West 8, Kita-ku, Sapporo, Hokkaido 060-8638, Japan; Global Station for Quantum Medical Science and Engineering, Global Institution for Collaborative Research and Education (GI-CoRE), Hokkaido University, North 14, West 5, Kita-ku, Sapporo, Hokkaido 060-8648, Japan. 2. Hokkaido University Hospital Clinical Research and Medical Innovation Center, Hokkaido University, North 14, West 5, Kita-ku, Sapporo, Hokkaido 060-8648, Japan. 3. Department of Diagnostic and Interventional Radiology, Hokkaido University Hospital, North14, West 5, Kita-ku, Sapporo, Hokkaido 060-8638, Japan. 4. Proton Beam Therapy Center, Hokkaido University Hospital, North14, West 5, Kita-ku, Sapporo, Hokkaido 060-8638, Japan. 5. Global Station for Quantum Medical Science and Engineering, Global Institution for Collaborative Research and Education (GI-CoRE), Hokkaido University, North 14, West 5, Kita-ku, Sapporo, Hokkaido 060-8648, Japan; Department of Radiation Medicine, Faculty of Medicine, Hokkaido University, North 15, West 7, Kita-ku, Sapporo, Hokkaido 060-8638, Japan. 6. Global Station for Quantum Medical Science and Engineering, Global Institution for Collaborative Research and Education (GI-CoRE), Hokkaido University, North 14, West 5, Kita-ku, Sapporo, Hokkaido 060-8648, Japan; Department of Radiation Medical Science and Engineering, Faculty of Medicine, Hokkaido University, North 15, West 7, Kita-ku, Sapporo, Hokkaido 060-8638, Japan.
Abstract
PURPOSE: To quantitatively evaluate and compare the image recognition performance of multiple fiducial markers available in real-time tumor-tracking radiation therapy (RTRT). METHODS: Clinically available markers including sphere shape, coil shape, cylinder shape, line shape, and ball shape (folded line shape) were evaluated in liver and lung models of RTRT. Maximum thickness of the polymethyl metacrylate (PMMA) phantom that could automatically recognize the marker was determined by template-pattern matching. Image registration accuracy of the fiducial marker was determined using liver RTRT model. Lung RTRT was mimicked with an anthropomorphic chest phantom and a one-dimensional motion stage in order to simulate marker motion in heterogeneous fluoroscopic images. The success or failure of marker tracking and image registration accuracy for the lung model were evaluated in the same manner as that for the liver model. RESULTS: All fiducial markers except for line shape and coil shape of thinner diameter were recognized by the PMMA phantom, which is assumed to have the typical thickness of an abdomen, with two-dimensional image registration accuracy of <2 pixels. Three-dimensional calculation error with the use of real-time stereoscopic fluoroscopy in RTRT was thought to be within 1 mm. In the evaluation using the lung model, the fiducial markers were recognized stably with sufficient accuracy for clinical application. The same was true for the evaluation using the liver model. CONCLUSIONS: The image recognition performance of fiducial markers was quantified and compared. The results presented here may be useful for the selection of fiducial markers.
PURPOSE: To quantitatively evaluate and compare the image recognition performance of multiple fiducial markers available in real-time tumor-tracking radiation therapy (RTRT). METHODS: Clinically available markers including sphere shape, coil shape, cylinder shape, line shape, and ball shape (folded line shape) were evaluated in liver and lung models of RTRT. Maximum thickness of the polymethyl metacrylate (PMMA) phantom that could automatically recognize the marker was determined by template-pattern matching. Image registration accuracy of the fiducial marker was determined using liver RTRT model. Lung RTRT was mimicked with an anthropomorphic chest phantom and a one-dimensional motion stage in order to simulate marker motion in heterogeneous fluoroscopic images. The success or failure of marker tracking and image registration accuracy for the lung model were evaluated in the same manner as that for the liver model. RESULTS: All fiducial markers except for line shape and coil shape of thinner diameter were recognized by the PMMA phantom, which is assumed to have the typical thickness of an abdomen, with two-dimensional image registration accuracy of <2 pixels. Three-dimensional calculation error with the use of real-time stereoscopic fluoroscopy in RTRT was thought to be within 1 mm. In the evaluation using the lung model, the fiducial markers were recognized stably with sufficient accuracy for clinical application. The same was true for the evaluation using the liver model. CONCLUSIONS: The image recognition performance of fiducial markers was quantified and compared. The results presented here may be useful for the selection of fiducial markers.