ePoster Forums
Purpose: To minimize the risk of brain necrosis (BN) in base-of-skull (BOS) patients treated with IMPT, we propose real-time per-voxel constraints (PVCs) to minimize hot spots in LET-guided robust optimization (LRO).
Methods: Ten BOS patients either having developed BN or having high risk to develop BN treated with IMPT were carefully selected. An optimizing structure (BrainOPT) and an evaluating structure (BrainROI) containing high risk regions were generated. Two plans were generated for every patient: conventional dose-only robust optimization (DRO) and LRO. PVCs were developed to control hot spots in either the dose, LET, or combined (dose/LET) distribution. Indices from dose-volume-histogram and extra-biological-dose (xBD)-volume histogram were used in the plan evaluation. Dose-LET-volume histogram (DLVH) was adopted to illustrate the underlying mechanism. Wilcoxon signed-rank test was used for statistical comparison.
Results: PVC effectively and efficiently minimized dose hot spots in both DRO and LRO and LET hot spots in LRO. PVC resulted in a non-stop workflow, which reduced the per “trial-and-error”iteration time during treatment planning from 13 minutes to 5 minutes. Compared to DRO, LRO could generate plans with statistically lower xBD hot spots in BrainROI with comparable target dose coverage, target dose hot spots, and dose hot spots in BrainROI. DLVH indicated that LRO could either reduce LET at the same dose level or redistribute high LET from high dose regions to low dose regions.
Conclusion: PVC is a powerful tool to minimize dose/LET hot spots in IMPT. LRO outperformed DRO in terms of controlling xBD hot spots.
Not Applicable / None Entered.