Automatic Generation of a Plan Optimization Volume for Tangential Field Breast Cancer Radiation Therapy |
| |
Authors: | Koen Van Vaerenbergh Werner De Gersem Luc Vakaet Marc Coghe Tom Boterberg Marlies Bakker Christina Derie Wouter Willaert Patricia Seij Wim Duthoy Carlos De Wagter Wilfried De Neve |
| |
Affiliation: | (1) Division of Radiotherapy, Ghent University Hospital, Gent, Belgium;(2) Radiotherapie en Kerngeneeskunde, De Pintelaan 185, 9000 Gent, Belgium |
| |
Abstract: | Background and Purpose: Dose homogeneity is one of the objectives during computer planning of postoperative radiotherapy of the conserved breast. For three-dimensional (3-D) optimization of the dose distribution using serial CT scan images, suitable volumes have to be delineated. The purpose of this study was to develop a computer-generated delineation of a plan optimization volume (POV) and an irradiated volume (IV) and to automate their use in a fast dose homogeneity optimization engine. Patients and Methods: Simulation was performed according to our standard procedure which involves the positioning of a lead collar around the palpable breast to facilitate the definition of gantry angle, collimator angle and field aperture for tangential wedged photon beams. In a change to the standard procedure an anterolateral radiograph was taken with its axis orthogonal to the central plane of the two tangential half-beams. Images from a serial CT scan were acquired in treatment position, and the geometric data of the three simulated beams were used by a computer program to generate the POV and IV. For each patient, weights of wedged and unwedged beams were optimized by either human heuristics using only the central slice (2-D), the whole set of CT slices (3-D), or by a computer algorithm using the POV, IV and lung volume with constrained matrix inversion (CMI) as optimization method. The resulting dose distributions were compared. Results: The total planning procedure took, on average, 44 min of which < 7 min were needed for human interactions, compared to about 52 min for the standard planning at Ghent University Hospital, Belgium. The simulation time is increased by 2–3 min. The method provides 3-D information of the dose distribution. Dose homogeneity and minimum dose inside the POV and maximum dose inside the IV were not significantly different for the three optimization techniques. Conclusion: This automated planning method is capable of replacing the contouring of the clinical target volume as well as the trial-and-error procedure of assigning weights of wedged and unwedged beams by an experienced planner. |
| |
Keywords: | Breast cancer Automated generation Plan optimization |
本文献已被 SpringerLink 等数据库收录! |
|