RESUMO
OBJECTIVE: To determine the overall tracking errors inherent to the co-calibration procedure of AlignRT InBore™'s (Vision RT Ltd., London, UK) ceiling-mounted and ring-mounted cameras. METHODS: Extrinsic calibration errors related to the mismatch between ceiling and InBore cameras' isocentres and treatment isocentre were determined using MV images and the SRS package and compared to traditional plate-based error. Next, using a realistic anthropomorphic female phantom, intrinsic calibration errors were determined while varying source-skin distance (80 to 100 cm), breast board inclination (0° to 12.5°), room lighting conditions (0 to 258 lx), skin colour (dark, white and natural skin colour), and pod occlusion. RESULTS: MV images of the cube proved plate-based calibration to suffer from large errors especially in the vertical direction (up to 2 mm). Intrinsic calibration errors were considerably lower. Indeed, RTD values of ceiling and InBore cameras showed little variability with isocentre depth (within 1.0 mm/0.4°), surface orientation and breast board inclination (within 0.7 mm/0.3°), changing lighting conditions (within 0.1 mm/0.2°), skin colour/tone (within 0.3 mm/0.3°) and camera pod occlusion (within 0.3 mm/0.2°). CONCLUSION: The use of MV-images proved critical to maintain co-calibrating errors of ceiling and InBore cameras to Halcyon's treatment isocentre below 1 mm.