23 First experience of multi-criteria optimization planning of tomotherapy plans

Abstract

Introduction Following the recent introduction of a Tomotherapy module in the RayStation system (RayTomo), we evaluated the capacities of the new multi-criteria optimizer (MCO), as well as its ability to deliver the plans with phantom measurements. Methods Fifteen patients (including head and neck, pelvis, and breast cases) previously treated with Tomotherapy Planning Station (TomoPS) from Accuray were re-planned with RayTomo. To avoid the variability and user dependence associated with a “classical” inverse optimization algorithm, MCO constraints were generated from a script taking into account the overlap of OARs with the targets [1] . Pareto plans were generated based on these constraints, and after achieving comparable target coverage in the plan navigation, the results were compared in terms of dose to organs-at-risk (OARs). Finally, a plan was exported to the machine and delivered on an Octavius phantom to obtain dose measurements with a 2D chamber array, and results were assessed using local gamma-analysis (3%-3 mm). Results Mean doses to important OARs were substantially reduced in head and neck and pelvic patients (Tables 1 and 2), while maximum doses remained essentially unchanged. Smaller mean dose reductions were also observed for the breast patients, together with a decrease in maximum doses. The Delivery Quality Assurance (DQA) plan showed a passing rate of ⩾ 95% with the 3%(local)-3 mm criterion routinely used in our center, indicating that the calculated MCO plan could reliably be delivered on the machine though the way to use the MLC is different. Download : Download high-res image (167KB) Download : Download full-size image Download : Download high-res image (94KB) Download : Download full-size image Conclusions Results obtained so far suggest that MCO with Tomotherapy might produce better results in OAR sparing than classical inverse optimization, but more importantly, it could greatly reduce the need for user input and interaction, and hence improve the homogeneity of the plans between users and institutions. Further dose measurements are ongoing to ensure the reliability of this new module.