Session Item

Automated treatment planning can increase plan quality and consistency while sparing resources, and has become a standard tool in many radiotherapy clinics. Erasmus-iCycle is a well-known system for automated planning, producing Pareto-optimal dose distributions according to a pre-defined wishlist.

Here we present a novel approach, where iCycle dose distributions are translated into objective templates for IMRT optimization in Eclipse. The method is validated on a set of locally advanced non-small cell lung cancer (LA-NSCLC) patients, comparing the deliverable iCycle-Eclipse (iC-E) plans to the manually created clinical plans (CP).

An iCycle wishlist for IMRT of LA-NSCLC was developed and tuned according to clinical priorities. An oncologist and a treatment planner were involved in this process. The order of priority for target and OAR objectives was PTV, lungs, heart and esophagus. In addition, there were hard constraints and objectives for normal tissue dose, and maximum dose constraints for the spinal canal, brachial plexus and PTV.

The iCycle DVHs were translated into patient-specific objective templates, with line objectives that limit dose for all volume levels for each OAR, and imported into Eclipse. Priorities and normal tissue objective settings were kept constant after tuning based on a few patients. The templates were used for automated plan optimization in Eclipse (no manual tuning), applying the same beam configurations as for the clinical plans, resulting in the iC-E plans.

iC-E plans for 17 LA-NSCLC patients were compared to the CP. All patients had a 6-beam, manually planned IMRT CP in Eclipse. The prescribed dose was 60 or 66 Gy in 2 Gy fractions and dose calculation was performed with Acuros XB 15.6. The Wilcoxon signed-rank test (p<0.05) was used for statistical testing.

Target coverage and lung dose were similar between CP and iC-E, while the average D_mean for heart and esophagus was significantly reduced with iC-E [Fig. 1a, Table 1]. iC-E reduced heart D_mean for 12/17 patients and esophagus D_mean for 13/17 patients [Fig. 2]. For the heart, it is noticeable that while the largest reductions with iC-E are for low doses, there is also a consistent decrease in the medium-high dose range [Fig. 1b].

The maximum dose to the spinal canal, brachial plexus and patient body followed the clinic’s requirements for all plans.

Table 1: Comparison of average dose metrics. Parameters in bold are significantly different.

The novel iCycle-Eclipse method for automated treatment planning was developed and validated on LA-NSCLC patients. A potential for sparing heart and esophagus dose for most patients compared to manual plans was revealed. The new method allows Eclipse to automatically generate plans that are close to Pareto-optimality.