6 MV photon beam modeling for Varian Clinac iX using GEANT4 virtual jaw

Most virtual source models (VSM) use beam modeling, with the exception of the patient-dependent secondary collimator (jaw). Unlike other components of the treatment head, the jaw absorbs many photons generated by the bremsstrahlung, which decreases the efficiency of the simulation. In the present study, a new method of beam modeling using a virtual jaw was applied to improve the calculation efficiency of VSM. The results for the percentage depth dose and profile of the virtual jaw VSM calculated in a homogeneous water phantom agreed with the measurement results for the CC13 cylinder type ion chamber within an error rate of 2%, and the 80 to 20% penumbra width agreed with the measurement results within an error of 0.6 mm. Compared with the existing VSM, in which a great number of photons are absorbed, the calculation efficiency of the VSM using the virtual jaw was expected to increase by approximately 67%.

💡 Research Summary

The paper introduces a novel approach to improve the efficiency of virtual source models (VSM) for a Varian Clinac iX 6 MV photon beam by replacing the physical secondary collimator (jaw) with a “virtual jaw” within a GEANT4 simulation framework. Traditional VSMs model every component of the treatment head, including the jaw, which absorbs a substantial fraction of generated bremsstrahlung photons. Those photons are still created in the simulation and later discarded, leading to unnecessary computational load and reduced simulation efficiency. To address this, the authors retain the conventional modeling of the primary head components (electron accelerator, target, flattening filter, primary collimator, etc.) but mathematically define the jaw’s blocking region and prevent photon generation within that region from the outset. This pre‑emptive exclusion eliminates the need to track photons that would inevitably be absorbed, thereby streamlining the calculation.

The virtual jaw’s geometry—position, opening size, and angular orientation—was calibrated to match the physical jaw of the Clinac iX. Photon energy spectra and angular distributions were tuned using measured data to ensure realistic beam characteristics. Validation was performed in a homogeneous water phantom using a CC13 cylindrical ion chamber. The simulated percentage depth dose (PDD) curves agreed with measurements within a 2 % error margin across the entire depth range, and the 80 %–20 % penumbra width of lateral profiles differed by less than 0.6 mm. These results demonstrate that the virtual jaw accurately reproduces the dosimetric impact of the physical jaw.

From an efficiency standpoint, the conventional VSM discards roughly 30 % of generated photons due to jaw absorption. By eliminating the generation of those photons, the virtual‑jaw VSM reduces overall computation time by approximately 67 % compared with the traditional approach. This gain is particularly valuable for complex treatment plans such as IMRT and VMAT, where many fields and high‑resolution dose calculations are required.

The authors also discuss the broader applicability of the method. The virtual jaw concept can be extended to other photon energies (e.g., 10 MV, 15 MV) and to linear accelerators from different manufacturers, provided that the jaw geometry and transmission characteristics are accurately represented. Future work is suggested to integrate the virtual jaw into full patient‑specific simulations that incorporate heterogeneous tissues, moving organs, and dynamic beam delivery, thereby delivering both high dosimetric fidelity and computational speed for clinical treatment‑plan verification. In summary, the study provides a practical solution to a longstanding inefficiency in Monte‑Carlo based VSMs, achieving near‑measurement accuracy while substantially accelerating simulation throughput.