A Multi-Institutional Model for EPID-Based 3D Dose Reconstruction in a Virtual Spherical Phantom for Benchmarking and Remote Auditing of Stereotactic Radiosurgery
Abstract
Purpose
For single-isocenter multi-target (SIMT) stereotactic radiosurgery (SRS), benchmarking, auditing and inter-institutional standardization of dose delivery remains challenging as it requires specialized equipment and expertise. Electronic portal imaging devices (EPIDs) offer a practical solution, where in-air images and EPID-to-dose conversions can be used for remote verification. This work aims to develop and validate an EPID-based technique to determine 3D dose in a virtual spherical phantom which is applicable across institutions for non-coplanar SIMT SRS dose verification.
Methods
Small-field output-factors (OFs) from 11 international centers were measured in-water for the TrueBeam linear accelerator to establish a reference dataset. This included jaw-defined and high-definition multi-leaf collimator (HDMLC) fields from 0.5x0.5 cm2 to 20x20 cm2 and depths from 1.5-20cm. Matching EPID images were also collected at each center. Utilizing the average-center data, two models (jaw-derived and MLC-derived) were created to convert cine-EPID images to 3D dose in a virtual spherical phantom. The model was validated at two centers by comparing EPID-derived dose with film in a 3D-printed spherical phantom for 11 non-coplanar SIMT SRS plans.
Results
Center-variation in the measured OFs for 0.5x0.5 cm2 fields at 10 cm depth was 6.9% and 2.1% (1SD) for the jaw-defined and MLC-defined fields respectively. For larger fields, variation was ≤0.8% (1SD). OFs estimated from EPID-to-dose models were within 1.8% (jaw-derived) and 1.0% (MLC-derived) of measured-in-water OFs at 10 cm depth. For SIMT SRS plans, target dose errors were -0.3±2.4% (mean±SD) and gamma pass rates were 99.6±1.7% (5%/1.5mm) for the jaw-derived model compared to film. For the MLC-derived model, dose errors of -0.2±2.1% and pass rates of 99.7±1.2% were achieved.
Conclusion
Variation in measured OFs across centres confirms that a single representative EPID-to-dose model can be utilized. The methodology has been independently validated and can be applied remotely for inter-institutional benchmarking, auditing and standardisation for SIMT SRS.