Improving plan quality in adaptive proton therapy using an interactive dose modification tool

Objective.Anatomical changes during proton therapy (PT) often necessitate plan adaptation. In online adaptation, re-optimizing using the original planning constraints has been proposed to speed-up the optimization time. However, the re-optimized plan may not reach the original plan quality. For this, an interactive dose modification (IDM) tool offers a promising approach, whereby fluences of proton pencil beams can be locally varied via interactive dose manipulations. We assessed the feasibility of using the IDM tool to improve the daily optimized plan, in cases where some clinical objectives are unmet.Approach.The IDM tool enables users to click on the dose distribution to locally increase/decrease dose, and underlying pencil beam weights, with immediate visual feedback. Four cases, treated with PT and exhibiting anatomical changes, were retrospectively studied in a simulated adaptive workflow (total 16 daily images analyzed). For each case, we compared three plans: a) re-calculated (but not re-optimized) on daily anatomy without adaptation, b) a plan re-optimized using the original planning constraints, c) a plan re-optimized as in b) but followed by modification using the IDM tool. We evaluated clinical endpoint fulfillment, dose deviations to clinical goals, and required modification time.Main results.Overall, plan a) resulted in 40 clinical endpoints being unmet that were fulfilled in the original plan, with deviations up to 9.4 Gy(RBE); plan b) resulted in 34 such endpoints, with deviations ⩽2.5 Gy(RBE); and plan c) reduced these to 3 and deviations to ⩽0.5 Gy(RBE). Target coverage (D98%, D95%) remained within 4 Gy(RBE) of the original plan for plan a) and 2.5 Gy(RBE) for plan b) and c). Average time for IDM was 2 min (30 s-6 min).Significance.The IDM tool enables efficient improvement of plan quality in adaptive PT. Its speed, intuitiveness and effectiveness in restoring clinical goals supports its potential integration into routine online adaptive workflows.