IMRT SX1+SX2 compare

Comparison of manual vs automatic IMRT optimization modes and 5mm effective MLC(SX2)+kV imaging vs 10mm MLC(SX1)+MV CBCT(10 MU)

Also avilable in xml template format in ZIP download below

The 2018 GYN ProKnow World Championship was a first of it’s kind four hour timed worldwide treatment planning competition.  A Halcyon nine field IMRT plan was submitted utilizing 5mm effective MLC mode (SX2) and utilizing kV CBCT imaging, features made available with the release of Halcyon 2.0+ and Eclipse 15.6+.  The submitted plan scored 135.37.

That submitted plan was created by running the Photon Optimizer in manual mode, allowing the cost function graph line to get completely flat before completing optimization, allowing the leaf motion calculator and dose calculation to complete before restarting the optimizer and using the current dose plan as intermediate dose for optimization and again letting the cost function line get flat before stopping optimization and letting the leaf motion calculator and dose calculator complete.  Any time during planning if an optimization objective was changed the complete optimization process would be started start over ensuring fully reproducible results by never changing optimization objectives during the optimization process.

Retrospectively, using the same “frozen in time” optimization objectives created for the originally submitted nine field IMRT plan, multiple plan variations were made.  Specifically, the difference between the optimization method above which lets the cost function line get flat and manually triggers an intermediate dose to ensure the Leaf Motion Calculation is accounted for in the intermediate dose optimization was compared to the automatic optimization with automatic intermediate dose, where the cost function line doesn’t get flat and the Leaf Motion Calculator isn’t run for the intermediate dose calculation.  Additionally, looking at the plan quality difference between the MLC in 5mm effective (SX2) mode versus the native 10mm mode (SX1), and seeing what effect, if any, could be seen in plan quality by not performing a 10MU MV CBCT on the patient each fraction and instead utilizing the kV imager for kV CBCT which does not contribute imaging dose accounted for by Eclipse in the treatment plan.

The below chart shows the raw score differences in these optimization/MLC/imaging plan variations for both IMRT and VMAT variations.  See the VMATv1+v2compare toggle at the top of this page for more information on the VMAT results.

Note: all of these scores could be made higher had the objectives been changed to values that were customized to maximize plan quality for those optimization settings.  The goal of this experiment was to quantify optimization differences while keeping objectives constant.  The ultimate highest scoring VMAT plan in this study (144.01) was made by Eclipse v13.7 user Jonathan Stenbeck manually pausing and interactively loading the optimizer.  A video walkthrough of his method can be found at www.proknowsystems.com after logging in and clicking the "learn" tab on this plan study.

_{Any reference to a "plan study" are simply what the organizers call each case and may not be a "study" in the FDA sense as they may not have been published in a peer reviewed journal.
Varian does not provide medical advice and these are illustrative examples only.
Leading plans by expert planner. Your results may vary.}