Transit dosimetry based on water equivalent path length measured with an amorphous silicon electronic portal imaging device

Abstract: Background and purpose: In vivo dosimetry is one of the quality assurance tools used in radiotherapy to monitor the dose delivered to the patient. The digital image format makes electronic portal imaging devices (EPIDs) good candidates for in vivo dosimetry. Currently there is no commercial transit dosimetry module, which could facilitate routine in vivo dosimetry with the EPID. Some centres are developing their in-house packages, and they are under assessment before introduction into routine clinical usage. The main purpose of this work was to develop the EPID as an in vivo dosimetry device. Materials and methods: Knowledge of a detector’s dose-response behaviour is a prerequisite for any clinical dosimetric application, hence in the first phase of the study, the dosimetric characteristics of eleven Varian a-Si500 EPIDs that are in clinical use in our centre were investigated. The devices have been in use for varying periods and interfaced with two different acquisition control software packages, IAS2 / IDU-II or IAS3 / IDU-20. Properties investigated include: linearity, reproducibility, signal uniformity, field size and dose-rate dependence, memory effects and image profiles as a function of dose. In the second phase, an EPID was calibrated using the quadratic method to yield values for the entrance and exit doses at the phantom or patient. EPID images for a set of solid water phantoms of varying thicknesses were acquired and the data fitted onto a quadratic equation, which relates the reduction in photon beam intensity to the attenuation coefficient and material thickness at a reference condition. The quadratic model was used to convert the measured grey scale value into water equivalent path length (EPL) at each pixel for any material imaged by the detector. For any other non-reference conditions, scatter, field size and MU variation effects on the image were corrected. The 2D EPL is linked to the percentage exit-dose for different thicknesses and field sizes, thereby converting the plane pixel values at each point into a 2D dose map at the exit surface of the imaged material. The off axis ratio is corrected using envelope and boundary profiles generated from the treatment planning system (TPS). The method was extended to include conformal and enhanced dynamic wedge (EDW) fields. A method was devised for the automatic calculation of areas (to establish the appropriate scatter correction) from the EPID image that facilitated the calculation of EPL for any field, and hence exit dose. For EDW fields, the fitting coefficients were modified by utilizing the Linac manufacturer’s golden segmented treatment tables (STT) methodology. Cross plane profiles and 2D dose distributions of EPID predicted doses were compared with those calculated with the Eclipse 8.6 treatment planning system (TPS) and those measured directly with a MapCHECK 2 device. Results: The image acquisition system influenced the dosimetric characteristics with the newer version (IAS3 with IDU-20) giving better data reproducibility and linearity fit than the older version (IAS2 with IDU-II). The irradiated field areas can be accurately determined from EPID images to within ± 1% uncertainty. The EPID predicted dose maps were compared with calculated doses from TPS at the exit. The gamma index at 3% dose difference (DD) and 3mm distance to agreement (DTA) resulted in an average of 97% acceptance for the square fields of 5, 10, 15 and 20 cm thickness solid water homogeneous phantoms. More than 90% of all points passed the gamma index acceptance criteria of 3% DD and 3mm DTA, for both conformal and EDW study cases. Comparison of the 2D EPID dose maps to those from TPS and MapCHECK shows that, more than 90% of all points passed the gamma index acceptance criteria of 3% dose difference and 3mm distance to agreement, for both conformal and EDW study cases. Conclusions: The quadratic calibration can effectively predict EPL and hence exit dose. Good agreement between the EPID predicted and TPS calculated dose distributions were obtained for open fields, conformal and EDW test cases. There were noteworthy deviations between EPID, TPS and MapCHECK doses on field edges. But it should be emphasised that, for practical in vivo dosimetry, these areas of reduced accuracy at the field edges are much less important. It is concluded that the EPID Quadratic Calibration Method (QCM) is an accurate and convenient method for online in vivo dosimetry and may therefore replace existing techniques