ESTRO 2024 - Abstract Book

S4756

Physics - Quality assurance and auditing

ESTRO 2024

444

Poster Discussion

Sensitivity and specificity evaluation of PerFRACTION in a HALCYON 2.0

Paula Navarro-Palomas, David Sánchez-Artuñedo

Hospital Universitari Vall d’Hebron, Servei de Física i Protecció Radiològica, Barcelona, Spain

Purpose/Objective:

PerFRACTION (Sun Nuclear Corporation) is a radiotherapy quality assurance software that monitors treatment fractions. PerFRACTION performs an independent dose calculation using the treatment logs with a collapsed cone algorithm. The PerFRACTION calculated dose distribution is compared with the treatment planning system dose distribution using different gamma criteria.

This work aims to evaluate the sensitivity and specificity of PerFRACTION under different known errors introduced in the treatment plans.

Material/Methods:

A head and neck, a gynecological, and a lung plan were retrospectively selected. All plans were calculated in Eclipse (Varian Medical Systems) with AAA (v.15.6) algorithm using a VMAT technique with 6 MV FFF photon beams for a Halcyon 2.0 linac. Following the SEAFARER methodology [1], known errors were introduced in the base plans using a Python script. Errors included variations in the MLC aperture (±0.1, ±0.2, ±0.5, ±1.0 and ±1.5 mm, and +0.2 and +0.5 mm only on the central leaves), collimator angle (±1°), monitor units (±2%, representing changes in the output), and combinations of them. The 17 modified plans for each base plan were recalculated in Eclipse. Errors were categorized based in the absolute change in the PTV mean dose: > 1%, > 2%, > 3% and > 5%. The modified plans were delivered in Halcyon to perform a fraction calculation in PerFRACTION (v2.10). For each modified plan, the difference in the PTV mean dose respect to the base plan was calculated. The PTV dose distribution for each modified plan was compared with the planned dose distribution using a global γ(3%/3 mm, TH (threshold) = 10%) in PerFRACTION. Experimental measurements of the modified and the base plans were carried out using ArcCHECK diode array (Sun Nuclear Corporation) and the linac Electronic Portal Imaging Device (EPID). Measurements were compared with the expected dose calculated from the base plan using a global γ(3%/3 mm, TH=10%). An acceptance gamma passing rate of 95% was established for PerFRACTION, EPID and ArcCHECK comparisons. The sensitivity and specificity of the three systems were studied. Receiver-operating characteristics curves (ROC curves) were constructed, and the area under curve (AUC) and the optimal cut-point passing rate value were obtained.