Evaluating robust optimisation for 4DCT in volumetric modulated arc therapy (VMAT) planning

Abstract

Background Setup, organ motion and density errors can cause delivered dose distributions to be different from planned. Lung treatments require planning methods to allow accurate tumour coverage when breathing motion is present. Traditionally, planning target volume (PTV) margins account for errors, however there remain inherent inaccuracies using these due to accounting for lung-tissue density heterogeneities for a mobile gross tumour volume (GTV). Robust optimisation offers a solution by including a full range of GTV positions within planning.

The aim of this research is to evaluate the difference between methods of robust optimisation for lung planning using 4DCT.

Methods

  • For a set of clinical VMAT lung patients with 4DCT scans:
    1. Optimise using the ITV on the 3D dataset with a 5mm PTV margin (currently used clinically).
    2. Optimise over the PTV on the average dataset.
    3. Outline GTV on all phases and optimise over every phase.
    4. Optimise using density override so the ITV is similar to the tumour density.
    5. Optimise using Raystation TPS with shift parameters based on patient data from a margin audit.
    6. Investigate plan quality sensitivity to target motion for each method by perturbing the dose distribution and importing 4D cone-beam CT scans into the planning system.
  • Programme an in-house phantom with realistic patient breathing traces obtained from planning 4DCT scans. 4DCT scan the phantom and create plans using the different robustness methods. Deliver these to the phantom and assess the dosimetric differences.
  • Extract patient breathing traces from patient 4D cone-beam CT scans, programme the in-house phantom with these and compare the dosimetry of robustness methods.
  • Investigate potential of uniform vs non-uniform dose delivery across the PTV to account for time spent in each phase. Potential impact of research Previous studies have used simplistic models to test robust optimisation methods rather than breathing traces from patients. This project is novel as it will dosimetrically test robust optimisation methods using realistic patient breathing traces. This may produce a different outcome to the simplistic models as a greater proportion of time is spent at the extremes of motion during patient breathing. Based on this assessment, optimisation of clinical plans using 4DCT may change if another approach is clearly less sensitive to target motion than the current margin technique. Patients will receive treatment where plan quality remains robust to target motion errors and hence will receive a higher quality treatment.

Outputs

Presented results so far at UKIO 2020. Will be presenting results at Raystation User Group Meeting 2020.