Erik van der Bijl1, René F M van Oers1, Igor Olaciregui-Ruiz2, Anton Mans1. 1. Department of Radiation Oncology, The Netherlands Cancer Institute - Antoni van Leeuwenhoek, Amsterdam, The Netherlands. 2. Department of Radiation Oncology, The Netherlands Cancer Institute - Antoni van Leeuwenhoek, Amsterdam, The Netherlands. Electronic address: i.olaciregui@nki.nl.
Abstract
BACKGROUND AND PURPOSE: To compare DVH-based quality assurance to a multi-parametric γ-based methodology for in vivo EPID dosimetry for VMAT to the pelvis. MATERIALS AND METHODS: For 47 rectum, 37 prostate, and 44 bladder VMAT treatments we reconstructed the 3D dose distributions of 387 fractions from in vivo EPID dosimetry. The difference between planned and measured dose was evaluated using γ analysis (3%/3mm) in the 50% isodose volume (IDV) and DVH differences (ΔD2, ΔD50 and ΔD98) of targets and organs at risk. The γ-indicators mean γ, γ pass rate and γ1% were compared to DVH-differences and their correlations were studied. DVH-based alerts on PTV and IDV were compared to γ-based alerts. RESULTS: Average PTV D50 and D98 dose differences were 0.0±2.2% (1SD) and -1.4±2.9% (1SD). Alert criteria of |ΔD50|<3.5-4.5% corresponded to an alert rate of about 10%. Strong correlations between mean γ and γ pass rate and difference in PTV ΔD50 were observed for all sites. DVH- and γ-based alerts agreed on >80% of the fractions for the majority of compared alert thresholds and methods. This agreement is >90% for the larger deviations. CONCLUSIONS: Strong correlations between some γ- and DVH indicators were found. Our comparison of multi-parametric alert strategies showed clinical equivalence for γ- and DVH-based methods.
BACKGROUND AND PURPOSE: To compare DVH-based quality assurance to a multi-parametric γ-based methodology for in vivo EPID dosimetry for VMAT to the pelvis. MATERIALS AND METHODS: For 47 rectum, 37 prostate, and 44 bladder VMAT treatments we reconstructed the 3D dose distributions of 387 fractions from in vivo EPID dosimetry. The difference between planned and measured dose was evaluated using γ analysis (3%/3mm) in the 50% isodose volume (IDV) and DVH differences (ΔD2, ΔD50 and ΔD98) of targets and organs at risk. The γ-indicators mean γ, γ pass rate and γ1% were compared to DVH-differences and their correlations were studied. DVH-based alerts on PTV and IDV were compared to γ-based alerts. RESULTS: Average PTV D50 and D98 dose differences were 0.0±2.2% (1SD) and -1.4±2.9% (1SD). Alert criteria of |ΔD50|<3.5-4.5% corresponded to an alert rate of about 10%. Strong correlations between mean γ and γ pass rate and difference in PTV ΔD50 were observed for all sites. DVH- and γ-based alerts agreed on >80% of the fractions for the majority of compared alert thresholds and methods. This agreement is >90% for the larger deviations. CONCLUSIONS: Strong correlations between some γ- and DVH indicators were found. Our comparison of multi-parametric alert strategies showed clinical equivalence for γ- and DVH-based methods.