Clinical implementation of a model-based in vivo dose verification system for SBRT-VMAT treatments using the EPID

Publication date: Available online 2 February 2017
Source:International Journal of Radiation Oncology*Biology*Physics
Author(s): Peter M. McCowan, Ganiyu Asuni, Eric Van Uytven, Timothy VanBeek, Boyd M.C. McCurdy, Shaun K. Loewen, Naseer Ahmed, Bashir Bashir, James B. Butler, Amitava Chowdhury, Arbind Dubey, Ahmet Leylek, Maged Nashed
PurposeElectronic portal imaging devices (EPIDs) are available on all modern linear accelerators. A system has been developed that uses on-treatment EPID images to reconstruct the in vivo dose delivered to the patient. We report findings from our in vivo dosimetry program implemented for all SBRT patients over a 31 month period, and discuss the value and challenges of utilizing in vivo EPID dosimetry clinically.MethodsFrom December 2013 to July 2016, 117 SBRT-VMAT patients (100 lung, 15 spine, and 2 liver) underwent 602 EPID-based in vivo dose verification events. A developed model-based dose reconstruction algorithm calculates the 3D dose distribution to the patient by back-projecting the primary fluence measured by the EPID during treatment. The EPID frame-averaging was optimized in June 2015. For each treatment, a 3%/3mm gamma (γ) comparison between our EPID derived dose and the Eclipse AcurosXB predicted dose to the planning target volume (PTV) and the ≥20% isodose volume were performed. Alert-levels were defined as γ pass-rates <85% (lung and liver) and <80% (spine). Investigations were carried out for all fractions exceeding the alert-level and were classified as: EPID-related, algorithmic, patient setup, anatomical change, or unknown/unidentified errors.ResultsThe percentages of fractions exceeding the alert levels were 22.6% for lung before frame-average optimization and 8.0% for lung, 20.0% for spine, and 10.0% for liver after frame-average optimization. Overall, mean PTV γ pass-rates were 90.7% ± 9.2%, 87.0% ± 9.3%, and 91.2% ± 3.4% for the lung, spine, and liver patients respectively.ConclusionsResults from the clinical implementation of our model-based in vivo dose verification method using on-treatment EPID images is reported. The method is demonstrated to be valuable for routine clinical use for verifying delivered dose as well as detecting errors.

Teaser

The results of a 31 month clinical implementation of our model-based, 3D patient in vivo dose verification system for all fractions of SBRT-VMAT treatments are given. Alert-levels based on 3D γ pass-rates in the PTV and ≥20% dose volumes were used. Flagged fractions were investigated to determine the cause of the discrepancy.


http://ift.tt/2kXBq2o