A first step towards EPID-based in-vivo dosimetry: distinguishing plan-specific from generic deviations

Patient-specific quality assurance (PSQA) is an essential step of radiotherapy workflow: it verifies that the delivered dose distribution coincides with the planned one, within tolerance levels. Electronic Portal Imaging Devices (EPIDs) have gained popularity because of their feature of reconstructing the delivered dose distribution inside patient's CT and the possibility to perform in-vivo dosimetry. In this thesis, carried out at AUSL-IRCCS Reggio Emilia, an EPID-based dosimetry system has been studied for pre-treatment verification. First, the calculation models were verified, to test Collapsed Cone (CC) algorithm and EPID reconstructions on an established protocol. The agreement found was satisfactory, indicating that the software for EPID dosimetry RadCalc, using CC, is reliable. Second, EPID performance on 40 treatment plans was compared with the one of the current standard for PSQA in Reggio Emilia, Octavius phantom; the aim was validating EPID for pre-treatment verification and finding a gamma criterion for EPID dosimetry able to represent Octavius behavior. The 3%/3mm local criterion, which is the reference on Octavius, was not considered suitable. When employing a 3%/2mm global normalization on both systems, a positive correlation was found, with small average bias. Third, to assess the sensitivity of both systems to delivery and calculation errors, the correlation between percentage dose differences (DDs) in the isocenter for the two systems and six complexity metrics was studied, in order to test Octavius and EPID sensitivity to the modulation of the same parameters. Octavius DDs showed statistically significant correlation with four metrics (EM, MCSv, ALG, PI), whereas for EPID no metric yielded a significant correlation. Finally, to exhibit EPID potential, a representative example of in-vivo dosimetry is shown. The gamma analysis of reconstructed distributions proved how EPID dosimetry can detect anatomical changes that require treatment replanning.