Room: AAPM ePoster Library
Purpose: To evaluate surface image registration for patient positioning in head and neck radiotherapy.
Methods: Surface image guided radiotherapy (SIGRT) and a Rando phantom were used for this study. The entire surface of the phantom’s face was covered with white masking tape in order to be visible by the SIGRT cameras. The phantom was immobilized using a standard clinical thermoplastic mask. The material from the midface portion of the mask was removed to provide an adequate surface for detection by SIGRT system. The phantom was CT scanned and a surface image was recorded using C-RAD Sentinel to create the reference image dataset. In the treatment room, both CBCT scan and surface image (using C-RAD CatalystHD) were acquired. After exporting image files from C-RAD, the 3D Normal Distribution Transform (NDT) point cloud registration algorithm was used to calculate the relative translations and rotations to align the treatment scan to the reference one. The shifts calculated by CBCT were compared with NDT registration results and Catalyst-calculated shifts.
Results: The average and standard deviation of difference between CBCT-calculated and point cloud registration shifts were -0.10 ± 0.40 mm translation and 0.0? ± 0.1 rotation along X, 0.22 ± 2.64 mm translation and 0.0? ± 0.1 rotation along Y, and -2.64 ± 0.15 mm translation and 0.4? ± 0.1 rotation along Z axes. The average and standard deviation of difference between CBCT-calculated and CatalystHD-calculated shifts were -0.32 ± 1.8 mm translation and -0.26? ± 0.58 rotation along X, 2.50 ± 2.52 mm translation and 0.46? ± 0.25 rotation along Y and -3.14 ± 0.37 mm translation and -0.46? ± 0.34 rotation along Z axes.
Conclusion: Surface image registration with an open mask is a viable option for positioning in head and neck radiotherapy, providing a high degree of registration accuracy.