Program Information
Investigation of ROI Dependency of a 3D Optical Surface Imaging System
Y Song*, C Obcemea, B Mueller, J Jeong, Z Saleh, X Tang, S Both, M Zinovoy, D Gelblum, B Mychalczak, Memorial Sloan-Kettering Cancer Center Westchester, West Harrison, NY
Presentations
SU-I-GPD-J-97 (Sunday, July 30, 2017) 3:00 PM - 6:00 PM Room: Exhibit Hall
Purpose: VMAT-based SRS often uses non-coplanar arcs with couch rotations greater than 20°. For these arcs, both CBCT and KV radiographic imaging cannot be safely performed for real-time patient monitoring. A viable alternative option is 3D optical surface imaging system (3D-OSIS). In this study, we investigated the interplay between the size of ROI and the detection accuracy of a commercial 3D-OSIS. Our goal was to establish an ROI threshold below which any ROI should not be used for image registration during clinical applications.
Methods: A circular phantom of 28.0 cm in diameter was constructed. To simulate a female breast, the central portion of the phantom was elevated by 3 cm. Prior to measurements, the phantom was placed on the couch and aligned with the machine isocenter. A reference surface image of the phantom was acquired and ROIs of 100%, 75%, 50%, and 25% of the phantom diameter were created for image registration. For each ROI, the couch was shifted by -1.0, -2.0, 1.0, and 2.0 cm in VRT, LAT, and LNG directions, respectively. For each couch shift, a static treatment surface image was acquired and registered with the reference surface image, which yielded 6-dimensional couch shifts.
Results: We found that the detection accuracy of 3D-OSIS was approximately linearly proportional to the size of ROI. As ROI became smaller, the detected shift differences became larger. When ROI was decreased to 50% of the phantom diameter, the detected shifts were off and became clinically unacceptable. When ROI was further decreased to 25% of the phantom diameter, the detected shifts were significantly off and meaningless.
Conclusion: Based on the results of our study, we strongly recommend that the ROI be at least 50% (in diameter) of the skin surface to be interrogated for reliable and accurate patient setup and real-time monitoring.
Contact Email: