Pre-clinical validation of a novel system for fully-automated treatment planning

Abstract

Introduction: Many approaches for automated treatment planning (autoplanning) have been proposed and investigated. Autoplanning can enhance plan quality compared to 'manual' trial-and-error planning, and decrease routine planning workload. A few approaches have been implemented in commercial treatment planning systems (TPSs). We performed a pre-clinical validation of a new system ('NovelATP') that is based on fully-automated multi-criterial optimization (MCO). The aim of NovelATP is to automatically generate for each patient a single high-quality, Pareto-optimal plan without manual Pareto navigation. Material and methods: Validation was performed by generating VMAT/IMRT plans for conventional treatment of prostate cancer (101 pts), prostate SBRT (20 pts), bilateral head-and-neck cancer (50 pts) and rectal cancer treated at an MR-Linac (23 pts). NovelATP autoplans were compared to plans that were generated with our in-house autoplanning system. In many previous validation studies, the latter system consistently showed enhanced plan quality when compared to manual planning. Results: Dosimetrical differences between NovelATP and benchmark plans were on average small and presumably not clinically relevant, pointing at high NovelATP dosimetric plan quality. MUs were 11-19% higher with NovelATP. NovelATP delivery times were up to 12% longer. Overall, there was a slight disadvantage for NovelATP regarding gamma analyses. Calculation times for NovelATP plans were between 29 and 151 min with no overall differences with the benchmark plans. Conclusion:The new autoplanning system was able to produce high-quality plans for four highly different planning protocols/treatment sites with a total of 194 patients investigated.

Automated Planning for Prostate Stereotactic Body Radiation Therapy on the 1.5 T MR-Linac

“… 4 Increased MUs and computed delivery times were also reported when comparing mCycle with its precursor, Erasmus-iCycle. 13 Such slight increases in computed MUs and delivery time from mCycle plans, as likely associated with an increased degree of modulation (but still reducible by tweaking of the segmentation parameters, if desired), was not associated with any significant degradation in deliverability metrics. All the in-phantom tested plans achieved a PR ≥ 90%, which is our minimum criterion for the clinical use of the plan.…”

“…The set of constraints and prioritized objectives for a specific treatment site and protocol defines a “wish list.” Applications of Erasmus-iCycle were reported for various anatomic sites such as head and neck, 6 , 7 cervix, 8 prostate, 9 and lungs. 10 More recently, Erasmus-iCycle was implemented into the Monaco TPS 11 , 12 , 13 as “mCycle,” the main novelty being the adoption of the physical and radiobiological cost functions of Monaco into the lexicographic logic.…”

Quality of life and toxicity guided treatment plan optimisation for head and neck cancer

“…Automated planning is nowadays a high-profile topic. Many different methods have been discussed in literature [11][12][13][14]. In most cases, a library of 'good' and 'representative' plans is available for learning the treatment planning system such that dose distributions can be predicted for new patients.…”

Treatment Planning in Intraoperative Radiation Therapy (IORT): Where Should We Go?

“…The same can be said for optimization strategies implemented in treatment planning systems: the relatively simple and standardized irradiation schemes in IORT did not push manufacturers to implement advanced optimization algorithms. For example, while in EBRT commercially available planning software includes multicriteria optimization and advanced autoplanning [ 11 , 12 ], direct planning is still largely in use in IORT.…”