SU‐E‐T‐197: Is the COMPASS 3D Dosimetry System An Accurate and Safe Quality Assurance Tool for Verifying Stereotactic Body Radiotherapy Dose Delivery for Non‐Small‐Cell Lung Carcinoma?

Purpose: Quality assurance for small field fractionated stereotactic radiotherapy treatment is critical and challenging due to the resolution of detectors used for measurements. We evaluate whether COMPASS system can be used as a good QA tool for SBRT dose delivery. Methods: 40 SBRT NSCLC cases were utilized in this study. PTV size ranged from 3cmx3.4cm to 8.1cmx6.7cm. Prescription doses to PTVs were 30Gy to 60Gy delivered in 3 to 5 fractions. For each case, a representative plan designed in Eclipse AAA using either RapidArc or dynamic‐IMRT was sent to COMPASS for calculating predicted dose. The reconstructed dose distributions were updated by using the measurement corrections. Measurements were made using a MatriXX Evaluation system mounted on the gantry at a SDD of 76.2cm, with a resolution of 7.62mm. The target dose‐volume parameters of D99%(dose received by 99% of target volume), D95%, D85%, D80%, D50%, D30%, D20%, D10%, D1%, average dose, and average gamma were analyzed against that of Eclipse. The OAR dose differences of the average dose and average gamma were also compared. Results: Compared to Eclipse, PTV average dose in COMPASS varied by 0.8% from predicated dose, and by 1.9% from reconstructed dose. PTV dose‐volume parameters show good agreement on predicated dose(±3%) and relatively poorer agreement on reconstructed dose(±7%). PTV dose‐volume parameters in low dose region show a better agreement than that in high dose region. The average gamma differences were up to 0.6 in PTV and 0.15 in OARs. The average absolute dose difference shows a maximum of 0.17Gy in OARs. There is no significant difference between VMAT and IMRT plans. Conclusion: COMPASS can be used as an independent QA tool for small field SBRT. COMPASS predicated doses show a better agreement with Eclipse when compared to reconstructed dose distributions. The limitation of the MatriXX resolution should be considered.