SU-E-T-291: Sensitivity of a Simple 2D EPID in Vivo Dosimetry




As radiotherapy (RT) increases in complexity, so does motivation for in vivo dosimetry (IVD), which may detect errors such as: setup, beam shaping and dose delivered. We have recently developed an easy-toimplement method for two-dimensional IVD based on images taken with the electronic portal imaging device (EPID) in cine mode during treatment. The purpose of this work is to characterize its sensitivity to possible RT delivery errors.


We introduced a series of modifications to a simple RT field (10×10, 100MU, 300RR, 20cm homogeneous phantom) to simulate errors. These modifications included multi-leaf collimator (MLC) position, number of MUs, and collimator angle. We quantified the sensitivity to inhomogeneities by inserting variable amounts of solid lung and bone. Finally we delivered realistic fields to an anthropomorphic phantom to estimate sensitivity to gantry angle and setup errors.


Our EPIDIVD is sensitive to MLC positioning errors of 1mm and 3mm in the closed and open directions respectively, and to 3% MU variations. Sensitivity to collimator angle depends on field shape irregularity; in the case of a 10×10 field, we are sensitive to errors of 0.8°. The sensitivity to inhomogeneities is limited by the nature of MV imaging: approximately 1% signal change is noted when switching 5cm of water to equal amounts of bone or lung. This suggests that the EPID-IVD is likely not sensitive to small setup or gantry angle errors, as confirmed by anthropomorphic tests.


We have characterized a simple method of 2D dose reconstruction at isocenter depth inside the patient, which is sensitive to possible RT delivery errors. This method may be useful as a secondary safety check, to prevent large errors from being carried on to following fractions, and to record delivered dose. By using readily available hardware, it is easily implemented and may prove especially useful in centers with limited resources.