28. Measurement of the improved plan quality consistency obtained using a knowledge-based treatment planning algorithm

Abstract

Purpose This study measured the reduction in human-caused variability driven by the use of RapidPlan (Varian Medical Systems, Palo Alto, CA). Methods Seven planners, with different levels of planning experience, were asked to deliver a VMAT treatment for fifteen prostate cancer patients with and without RapidPlan assistance. The plans were compared on the basis of target coverage, conformance and sparing of rectum, bladder and femnoral heads. The Plan Quality Metric formalism was used as an overall measure of plan quality. Inter-planner and intra-planner variability were measured in terms of differences in mean values and InterQuartile Ranges patients and operators. Results Target coverage, homogeneity, conformance and bladder sparing were similar between manual and RapidPlan assisted plans. But, the latter showed increased sparing of rectum and femoral heads sparing. 8 out of 15 patients showed a statistically significant increase in overall quality, as measured by percentage PQM%. RapidPlan assistance induces a reduction in inter-planner variability for rectum and femoral heads (p-value RapidPlan assistance induced a reduction of the variability of the overall plan quality, as a result of a large increase in plan quality for the less experienced planners. In general the inter-planner, IQR of PQM%, was reduced from 8.32 ± 4.19 to 4.73 ± 3.79 (p-value = 0.0046). At the same time, a reduction in intra-planner variability is measured with a significant overall reduction (p-value = 0.033). Conclusions The assistance of RapidPlan during the optimization of prostate cancer treatments induces a significant increase of plan quality and a contextual reduction of plan variability. RapidPlan is proven to be a valuable tool to leverage the planning skills of less experienced planners ensuring a better homogeneity of treatment plan quality.