Purpose: During online adaptive planning with the Elekta Unity MR-Linac, new intensity modulated radiotherapy (IMRT) treatment plans are created for each patient immediately prior to treatment. To reduce both offline and online planning time, Elekta has developed a new Optimal Fluence Levels leaf sequencer and pseudo-gradient descent shape optimizer (OFL+PGD) in a research version of the Monaco treatment planning system.
Methods: Unity MR-Linac Step-and-shoot IMRT beam complexity metrics were calculated for plans generated using Elekta’s research OFL+PGD algorithms and Elekta’s current optimizer and leaf sequencer (Hyperion). A total of 10 patients were included in this preliminary study (3 prostate, 2 pancreas, 3 oligomets, and 2 liver). Beam complexity metrics include plan averaged beam area normalized to PTV area (NPA), plan averaged beam irregularity (PI), plan normalized MU (PMU), and plan averaged beam modulation (PM). Total number of segments, and number of MLC subsegments (i.e. separated open areas within a segment) was also quantified. Statistical significance between the OFL+PGD and Hyperion algorithms results was assessed via student t-test and correlations through Spearman’s rank test.
Results: Differences in IMRT beam complexity metrics between OFL+PGD and Hyperion algorithms were determined as follows: ΔNPA=-0.004±0.025, ΔPI=-0.2±0.5, ΔPMU=-37.0±59.4 MU, and ΔPM=-0.1±0.1. OFL+PGD algorithms also produced fewer segments and subsegments (-10.1±9.7 and -0.2±0.2 respectively) compared to the Hyperion algorithm. The pairwise reduction in PM, number of segments, and number of subsegments was statistically significant (p=0.004,p=0.009,p=0.02 respectively). Additionally, PMU was found to be correlated to PM for both the OFL+PGD and Hyperion algorithms.
Conclusion: For the 10 plans evaluated, the new OFL+PGD algorithm produced plans with reduced PM, number of segments, and number of subsegments compared with Hyperion. Differences in NPA, PI and PMU were not found to be statistically different between the two algorithms. For both Hyperion and OFL+PGD algorithms, PMU was correlated with PM.
Funding Support, Disclosures, and Conflict of Interest: Funding for this work was supplied through an industry grant from Elekta.