Consistency of absorbed dose to water measurements using 21 ion-chamber models following the AAPM TG51 and TG21 calibration protocols

Abstract

In 1999, the AAPM introduced a reference dosimetry protocol, known as TG51, based on an absorbed dose standard. This replaced the previous protocol, known as TG21, which was based on an air kerma standard. A significant body of literature has emerged discussing the improved accuracy and robustness of the absorbed dose standard, and quantifying the changes in baseline dosimetry with the introduction of the absorbed dose protocol. A significant component playing a role in the overall accuracy of beam output determination is the variability due to the use of different dosimeters. This issue, not adequately addressed in the past, is the focus of the present study. This work provides a comparison of absorbed dose determinations using 21 different makes and models of ion chambers for low- and high-energy photon and electron beams. The study included 13 models of cylindrical ion chambers and eight models of plane-parallel chambers. A high degree of precision (<0.25%) resulted from measurements with all chambers in a single setting, a sufficient number of repeat readings, and the use of high quality ion chambers as external monitors. Cylindrical chambers in photon beams show an improvement in chamber-to-chamber consistency with TG51. For electron dosimetry with plane-parallel chambers, the parameters Ngas and the product ND,w x k(ecal) were each determined in two ways, based on (i) an ADCL calibration, and (ii) a cross comparison with an ADCL-calibrated cylindrical chamber in a high-energy electron beam. Plane-parallel chamber results, therefore, are presented for both methods of chamber calibration. Our electron results with technique (i) show that plane-parallel chambers, as a group, overestimate the beam output relative to cylindrical chambers by 1%-2% with either protocol. Technique (ii), by definition, normalizes the plane-parallel results to the cylindrical results. In all cases, the maximum spread in output from the various cylindrical chambers is <2% implying a standard deviation of less than 0.5%. For plane-parallel chambers, the maximum spread is somewhat larger, up to 3%. A few chambers have been identified as outliers.