Accurate unfolding of photon spectra of clinical linear accelerators using transmission measurements
Abstract
Most cancer radiation therapy treatments use external photon beams from clinical linear accelerators. For these beams, transmission analysis is a potentially viable approach for independent determination of the photon spectra and the incident electron energies. This study presents a comprehensive physics-based approach to transmission analysis to address the computational and experimental limitations of previous studies. On the computational side, energy differentiation is significantly improved by using transmission data from an optimum combination of multiple attenuators and detectors. Detector energy response, photonuclear attenuation, and corrections for non-ideal exponential attenuation are accounted for and found to have a major effect on the unfolding accuracy. For robust unfolding, the spectra are specified using a new, validated functional form with four free parameters, one of which is the incident electron energy. On the experimental side, the validation is performed on a research linac whose photon spectra and electron beam parameters are directly and independently known. The validation includes eight beams from 10 to 30 MV, with thick bremsstrahlung targets of Be, Al, and Pb. The approach is demonstrated on a clinical linac for 6, 10 and 25 MV beams. A protocol is developed to account for many experimental influence quantities, allowing for measurement accuracy of 0.4% on the smallest signals. The unfolded spectra agree with the benchmark spectra with root-mean-square energy fluence deviations of 4.5%. The accuracy of unfolding the incident electron energy is shown to be 3%. The overall accuracy improvement over the best previous studies is at least a factor of 3. Photon cross section uncertainties are the ultimate limiting factor of the technique. An upper bound estimate at the 95% confidence level for these uncertainties is found to be 0.7%, which is more realistic than the currently used 'envelope of uncertainty' of 1 -- 2%. By-products of this study include benchmarking the EGSnrc Monte Carlo system for relative ion chamber response calculations at the 0.2% level, and upgrading EGSnrc to model photonuclear attenuation.
- Publication:
-
Ph.D. Thesis
- Pub Date:
- 2012
- Bibcode:
- 2012PhDT.......250A
- Keywords:
-
- Physics, Radiation