Room: Davidson Ballroom A
Purpose: The use of convolution superposition for patient dose calculations relies on the knowledge of the primary energy spectrum of a treatment linear accelerator. This spectrum is determined during commissioning by unfolding data from percent depth dose (PDD) curves. This method is indirect and insensitive to small spectral variations. The most accurate form of spectral determination is through direct measurement using a pulse mode detector, which has previously been unsuccessful due to the high-fluence, high-energy nature of linear accelerator photon beams. This work optimized spectrometry techniques to measure the energy spectrum of a 6 MV linear accelerator and to compare the measured spectrum with Monte Carlo simulations.
Methods: Spectra were measured using a high purity germanium spectrometer with a Varian Clinac 21 EX linear accelerator. Two methods were implemented to reduce fluence to the detector, including a custom-built lead shield and implementing a Compton Scattering (CS) spectrometry setup. The detector was placed at CS angles of 125 and 89 degrees from the central beam axis and a response function was generated for each CS angle to account for photon interactions within the experimental geometry. After correcting the raw spectrum with a response function, Goldâ€™s deconvolution was used to unfold the energy spectrum. The measured spectrum was compared with MCNP6 Monte Carlo simulations, and an accelerator geometry that was experimentally validated by comparing measured and simulated PDDs and cross-field profiles.
Results: Comparisons showed good agreement between, (1) measured spectra at different CS angles corrected back to the primary spectrum, and (2) measured and simulated spectra. All points above 50 keV were within the estimated uncertainty of the measurement.
Conclusion: This work demonstrated that the direct measurement of the energy spectrum is feasible and showed good agreement with simulations.