Averaging VMAT treatment plans for multi-criteria navigation

TL;DR

This paper demonstrates that averaging leaf positions in VMAT plans can produce dose distributions close to the average of original plans, enabling real-time multi-criteria treatment plan navigation.

Contribution

It introduces a method to average VMAT plans by leaf positions, facilitating interactive Pareto surface navigation in radiation therapy planning.

Findings

Averaged leaf positions produce approximately averaged dose distributions.

The method enables real-time multi-criteria plan exploration.

Applicable to both photon and proton therapy plans.

Abstract

The main approach to smooth Pareto surface navigation for radiation therapy multi-criteria treatment planning involves taking real-time averages of pre-computed treatment plans. In fluence-based treatment planning, fluence maps themselves can be averaged, which leads to the dose distributions being averaged due to the linear relationship between fluence and dose. This works for fluence-based photon plans and proton spot scanning plans. In this technical note, we show that two or more sliding window volumetric modulated arc therapy (VMAT) plans can be combined by averaging leaf positions in a certain way, and we demonstrate that the resulting dose distribution for the averaged plan is approximately the average of the dose distributions of the original plans. This leads to the ability to do Pareto surface navigation, i.e. interactive multi-criteria exploration of VMAT plan dosimetric tradeoffs.