Encrypted login | home

Program Information

Consistency of Multi-Institutional Quality Assurance for Nasopharynx Volumetric Arc Therapy (VMAT)

no image available
D Thwaites

D Thwaites1*, E Pogson2 , S Arumugam3 , C Hansen4 , M currie5 , S Blake6 , N Roberts7 , M Carolan8 , P Vial9 , J Juresic10 , J Yakobi11 , A Haman12 , A Trtovac13 , L Holloway14 , (1) University of Sydney, Camperdown, Sydney, ,(2) University of Sydney, Sydney, NSW, (3) Liverpool and Macarthur Cancer Therapy Centres, Liverpool, NSW, (4) Odense University Hospital, Odense, OD, (5) Illawarra Cancer Care Centre, Wollongong, NSW, (6) University of Sydney, Sydney, NSW, (7) University of Wollongong, Wollongong, NSW, (8) Illawarra Cancer Care Centre, Wollongong, NSW, (9) Liverpool Hospital, Liverpool, NSW, (10) Liverpool Hospital, Liverpool, NSW, (11) Liverpool Hospital, Liverpool, NSW, (12) Liverpool Hospital, Liverpool, NSW, (13) Liverpool Hospital, Liverpool, NSW, (14) Liverpool & Macathur Cancer Therapy Centre, Sydney, New south Wales


SU-I-GPD-T-275 (Sunday, July 30, 2017) 3:00 PM - 6:00 PM Room: Exhibit Hall

Purpose: Quality assurance of individual treatment plan delivery accuracy is often performed using a dosimeter/phantom combination and gamma analysis threshold criteria. Collapsing differences for a complex treatment plan into a single number is problematic. This study evaluates the sensitivity of different equipment, in different planning approaches, to simulated machine errors.

Methods: Up to 10 patient VMAT plans were generated in Pinnacle3 at three institutions, following each's usual protocol, including auto-planning in one centre. Machine errors were introduced utilising Python, including collimator (C,°), MLC field size (F,mm) and MLC shift (S,mm) errors of ±5, ±2, ±1 units. Error plans were recalculated and were considered clinically unacceptable if any DVH metric differences relative to the baseline plan were above the following tolerances: any PTV D95%, brainstem or spinal cord D1cc, ≥±5%; parotids Dmean, ≥±10%. The smallest error plans deemed unacceptable for each error type were delivered and measured (Elekta Linac and Arccheck in two, Varian Linac and Arccheck and Delta4 in one). Gamma analysis was performed in SNC Patient v6.6 or Delta4 software with 3%/3mm and 2%/2mm global criteria (10% threshold, correction off). Before measurements, benchmarking tests were carried out to assure linac performance.

Results: The global 3%/3mm gamma pass detected the majority of unacceptable plans, although some still passed, considering gamma pass rates of < 95% (for 3%/3mm) or <88% (2%/2mm) as failing. The error type that passed was different at different institutions, although some MLC-F error plans passed using the Arccheck in all institutions. The auto-plans had similar pass rates to manual plans, despite the higher error magnitude in the latter.

Conclusion: Not all clinically-unacceptable introduced errors were discovered for VMAT plans using a typical 3%/3mm gamma criteria. Consistency was low for plans assessed utilising differing devices and software. A 2%/2mm global analysis was more sensitive to the errors considered.

Contact Email: