TH-CD-304-01: Evaluating the Accuracy of Absolute Portal Dosimetry On the EDGE Linear Accelerator

Abstract

Purpose: To validate the accuracy of the Varian (Palo Alto, CA) absolute portal dosimetry (APD) platform for the EDGE linear accelerator and evaluate the performance of an improved, non-transit EPID dosimetry system which is optimized to mitigate the known limitations of the current, clinically available solution. Methods: Measurements were performed using the Varian EDGE in 6X FFF mode and aS1200 MV detector—a new, high energy, backscatter-shielded, amorphous silicon EPID (43×43 cm 2 ). Integrated images were acquired for 124 clinical, IMRT/VMAT fields. The raw images were corrected for pixel sensitivity variation and off-axis differential energy response and converted to dose in water at 5cm depth using an empirical dose model. Delivered doses were compared to reference doses calculated in a water phantom using the TPS dose calculation algorithm. Per-beam analysis was performed using gamma evaluation with 3%/1mm dose difference/distance-to-agreement specifications. Additionally, composite EPID doses were calculated for 16 IMRT/VMAT treatment plans and benchmarked against gafchromic EBT3 film measurements, with agreement being analyzed using 3%/3mm, 2%/2mm and 3%/1mm gamma criteria. Results: The converted EPID doses showed good agreement with calculated doses to water. An average of 98.1% (± 4.4%) of pixels passed per-beam gamma analysis with a stringent 3%/1mm criteria. The comparison of EPID measurements with film also yielded high pass rates for varying plan complexities and over a wide range of doses (maximum planar dose: 2.6–36.4 Gy) for each level of gamma criteria. Average gamma indices were 99.7%, 96.1% and 90.3% for 3%/3mm, 2%/2mm and 3%/1mm parameters, respectively. Conclusion: The APD platform on the EDGE includes hardware upgrades, advanced detector calibration techniques and more sophisticated calculation capabilities which allow for accurate, high-resolution absolute dose verification for conventional as well as SRS and SBRT IMRT/VMAT cases. Further investigation is needed to evaluate the error detectability of this QA method. This work was supported in part by Varian Medical Systems (Palo Alto, CA)