|
内容記述 |
Purpose: In proton therapy, if the proton range and dose distribution in a patient's body differ from those calculated in the treatment planning, not only is the anti-tumor effect reduced, but normal tissue damage can increase. A planar PET system has been developed to visualize a distribution of annihilation gamma-ray emissions (activity distribution) correlated to the proton irradiation volume. Recently,“ pre-irradiation” procedure is proposed to verify in vivo dose delivery; pencil beams are irradiated at multiple spots just before a main treatment and the range positions are confirmed from the obtained PET images. In this procedure, the irradiation dose and measurement time directly influence the accuracy of range estimation. This study presents a method for predetermining the irradiation dose and its corresponding optimum measurement time (i.e., minimum) to achieve accurate range estimation.Methods: Assuming that the statistical noise in PET measurements obeys the Poisson distribution, the range uncertainties originated from the number of counts and the magnitude of gradient can be estimated by the Fisher's information. In addition, the range difference determined from simulated (i.e., ideal) and measured activity distributions can be estimated by the maximum likelihood estimation (MLE) algorithm. In this study, 100.2-MeV proton beams at the peak doses equivalent to 0.1, 0.5, and 2 Gy were irradiated on polyethylene targets, and the PET measurements were conducted for 300 s soon after the irradiations in both Monte Carlo simulation (using Geant4-v11.2.1) and experiment (conducted at Nagoya Proton Therapy Center); the 1D activity distributions integrated along the depth direction were obtained. From the simulated data, the measurement times when the range uncertainties were expected within 1 mm at 2σ were determined by the Fisher's information. Then, the actual range differences between the simulations and measurements were obtained by the MLE algorithm.Results: From the Fisher's information based on the simulated data at the peak doses of 0.1, 0.5, and 2 Gy, the range uncertainties were expected within 1 mm at 2σ when the measurement times were 120, 30, and 20 s, respectively. At these measurement times, the actual range differences observed in the measured data were 0.43, 0.44, and 0.27 mm, respectively. The range differences obtained in the actual measurements were consistent with those estimated by the Fisher's information.Conclusion: The results demonstrated the validity of applying the Fisher's information to determining the minimum dose and measurement time for accurate range estimation using PET in proton therapy. |
