Development and clinical implementation of a simple knowledge-based planning tool for prostate volumetric modulated arc therapy

Abstract

The aim of this work was to develop and incorporate a simple knowledge-based planning (KBP) tool into the volumetric modulated arc therapy (VMAT) treatment plan production process and to determine whether the tool improved plan quality and consistency. A study of 92 existing clinical prostate VMAT plans showed correlation between the proportion of organ at risk (OAR) overlapping with the planning target volume (PTV) and various OAR volume-doses of clinical interest. Simple linear models were generated that predicted the achievable rectum and bladder volume-doses according to the fraction of OAR volume overlapping with the PTV. The models were integrated into the planning process via a script using the Eclipse scripting application programming interface (ESAPI). The script’s impact on plan quality was evaluated by comparing the plans before and after script implementation using the parameter d = achieved volume-dose—predicted volume-dose, the amount by which the achieved volume-dose exceeded the predicted value. All OAR volume-doses investigated demonstrated smaller d and reduced variability following script implementation. The largest reductions in d value were observed for bladder V60Gy and V50Gy, which reduced from 2.9 ± 2.3 to 0.5 ± 1.6, and 4.9 ± 4.7 to 1.5 ± 3.4 respectively. Variability was most significantly improved for rectum V50Gy, for which the standard deviation was reduced from 7.2 to 3.8% following script implementation. A survey of the treatment planners found that the majority believed the script helped to improve the quality and consistency of their plans (57%) and save time (86%).