Development of robustness evaluation strategies for enabling statistically consistent reporting.

Area of Science:

• Medical Physics
• Radiation Oncology
• Computational Biology

Background:

• Proton therapy plan robustness evaluation is vital for safe treatment delivery.
• Current methods often lack comprehensive error exploration and confidence quantification.

Purpose of the Study:

• To compare established and advanced robustness evaluation procedures for proton therapy.
• To quantify confidence levels associated with different evaluation methods.

Main Methods:

• Evaluated methods included Good Practice Scenario Selection (GPSS) and Statistically Sound Scenario Selection (SSSS).
• Statistically Sound Dosimetric Selection (SSDS) was assessed using D95 (SSDS_D95) and objective function approximation (SSDS_OF).
• Simulations used MCsquare with systematic/random setup and range errors for 5 head-and-neck patients.

Main Results:

• SSSS methods produced similar results and comparable target coverage to GPSS.
• SSDS_D95 and SSDS_OF showed larger worst-case D98 for the CTV compared to GPSS.
• Explicitly simulating random errors in SSDS improved DVH metrics and plan robustness insights.

Conclusions:

• The choice of scenario sampling method impacts DVH-band width and confidence levels.
• Statistically sound dosimetric evaluation provides superior insight into proton therapy plan robustness.