A Comparison of In-Depth Ion Chamber Measurements and Epid Response for Enhanced Leaf Model Configuration
Abstract
Purpose
To investigate the effects of the Varian enhanced leaf model (ELM) on portal dosimetry.
Methods
Transmission factors (TR) and sweeping gap were measured for five energies at 10 cm depth using the ELM pattern plans. Plans were delivered on a TrueBeam LINAC (HD-MLC) using a Farmer chamber. These measurements were used to configure the ELM, which is used in portal dose image prediction (PDIP). The plans were delivered to the EPID and portal measurements were extracted from a 7×7mm2 central region for comparison against ion chamber baselines.
Results
PDIP estimates matched ion chamber measurements within 0.1% across all energies. However, portal measurements deviated from baselines depending on energy. TR comparisons (ion chamber vs. portal) were: 6X (1.10% vs 0.89%), 6FFF (0.95% vs 0.53%), 10X (1.25% vs 1.15%), 10FFF (1.12% vs 0.86%), and 15X (1.20% vs 1.18%). Sweeping gap measurements followed a similar trend, where discrepancies were largest for lower energy and FFF beams (>0.59% for 6FFF vs. <0.06% for 15X). Qualitatively, the ELM correctly models lower transmission at the leaf tip relative to the drive screw region but fails to capture a distinct transmission dip at the screw end visible in portal images. Additionally, PDIP extensive smoothing obscures the tongue-and-groove effect seen in portal images.
Conclusion
We demonstrated that portal-measured TR differs from the ELM baseline values, due to measurement depth differences and MLC-induced beam hardening. This discrepancy is most prominent in 6FFF beam, where portal transmission is only 56% of the modeled transmission. For multi-target SRS plans where targets are blocked by MLC for the majority of beam-on time, this discrepancy can introduce a systematic error in QA results. Caution is advised when interpreting portal QA for highly modulated FFF plans, as disagreement may stem from detector response modeling rather than delivery inaccuracy.