Eclipse 10

The beam angles of IMRT were initially optimized by the beam angle optimization algorithm (Varian Eclipse 10.0); a set of initial optimization objectives were loaded into the treatment planning system. The number of the fields was confined to five. Some beam angles were adjusted according to the experience of the dosimetrists, if the results of the beam angle optimization did not satisfy the dosimetric criteria. The plans were iteratively optimized to obtain the optimal PTV coverage and OARs sparing. After inverse planning, the leaf sequences using sliding window technique were generated for IMRT plans.

OARs including the parotid gland (PG), the submandibular gland (SMG), the cervical vertebra (VTB) and the vertebral foramen (VF), on both PCT and CBCT were manually delineated by an experienced physician using a commercial treatment planning system (TPS, Eclipse 10.0, Varian) and double checked by this same physician three months later to ease intra-observer variations. The contours delineated on the PCT served as the moving contours being deformed to the reference fractional CBCT domain, while the contours delineated on the CBCT images served as the ground truths for contour propagation accuracy validation in this study. We chose the deformable organ PG and SMG instead of the gross target volume (GTV) for this evaluation study for two reasons. First, many pilot studies have revealed that large volume shrinkage or deformation can be seen in the PG and the SMG during H&N radiation treatment [4 (link), 25 (link)–36 (link)], and thus, contour propagation of the PG and SMG onto the fractional CBCT anatomy is usually necessary for treatment re-planning or planning optimization. Second, GTV in CBCT images is difficult to identify due to the low soft tissue contrast and severe image artifacts present in CBCT. Using GTV with erroneous contouring in CBCT images as the ground truth would bias the evaluation result.

The Hybrid IMRT/VMAT plans were combination of 1-full-arc VMAT (Hybrid-VMAT) and 7-field IMRT (Hybrid-IMRT). The Hybrid-IMRT plans delivered half of the prescribed dose while the Hybrid-VMAT parts consisted of one full arc which was optimized with the Hybrid-IMRT plan as a base plan, to deliver the other half prescribed dose. The beam angles of Hybrid-IMRT were initially optimized by the beam angle optimization (BAO) algorithm (Varian Eclipse 10.0). The number of the fields was confined to seven. Some beam angles were adjusted according to the experience of the dosimetrist, if the results of the BAO did not satisfy the dosimetric criteria. The same optimization objectives and planning parameters were used for the Hybrid IMRT/VMAT, 2ARC-VMAT, and 9F-IMRT plans.