Automatic replanning of VMAT plans for different treatment machines: A template-based approach using constrained optimization

• Published in: Strahlentherapie und Onkologie 2018-10, Vol.194 (10), p.921-928
• Main Authors: Künzel, Luise A., Dohm, Oliver S., Alber, Markus, Zips, Daniel, Thorwarth, Daniela
• Format: Article
• Language: English
• Subjects: Cancer; Constraints; Medicine; Medicine & Public Health; Oncology; Optimization; Original Article; Prostate; Prostate cancer; Radiation therapy; Radiotherapy; Therapy
• ISSN: 0179-7158; 1439-099X
• DOI: 10.1007/s00066-018-1319-x

Purpose To investigate a new automatic template-based replanning approach combined with constrained optimization, which may be highly useful for a rapid plan transfer for planned or unplanned machine breakdowns. This approach was tested for prostate cancer (PC) and head-and-neck cancer (HNC) cases. Methods The constraints of a previously optimized volumetric modulated arc therapy (VMAT) plan were used as a template for automatic plan reoptimization for different accelerator head models. All plans were generated using the treatment planning system (TPS) Hyperion. Automatic replanning was performed for 16 PC cases, initially planned for MLC1 (4 mm MLC) and reoptimized for MLC2 (5 mm) and MLC3 (10 mm) and for 19 HNC cases, replanned from MLC2 to MLC3. EUD , D mean , D 2% , and D 98% were evaluated for targets; for OARs EUD and D 2% were analyzed. Replanning was considered successful if both plans fulfilled equal constraints. Results All prostate cases were successfully replanned. The mean relative target EUD deviation was −0.15% and −0.57% for replanning to MLC2 and MLC3, respectively. OAR sparing was successful in all cases. Replanning of HNC cases from MLC2 to MLC3 was successful in 16/19 patients with a mean decrease of −0.64% in PTV60 EUD . In three cases target doses were substantially decreased by up to −2.58% (PTV60) and −3.44% (PTV54), respectively. Nevertheless, OAR sparing was always achieved as planned. Conclusions Automatic replanning of VMAT plans for a different treatment machine by using pre-existing constraints as a template for a reoptimization is feasible and successful in terms of equal constraints.