EPID Dosimetry Comparisons and Corrections: Portal Image Prediction Vs. Portal Dose Conversion
D Bailey1*, L Kumaraswamy1, M Bakhtiari2, M Podgorsak1, (1) Roswell Park Cancer Institute, Buffalo, NY, (2) RadAmerica, LLC--MedStar Health, Baltimore, MDWE-G-108-7 Wednesday 4:30PM - 6:00PM Room: 108
Purpose: To compare the pretreatment quality assurance robustness of two forms of EPID dosimetry: dose image prediction and dose plane conversion. Both systems are tested for accuracy in all areas of the detecting surface, for both symmetric and asymmetric field sizes, and various levels of modulation intensity. Both EPID dosimetry techniques are sensitivity-tested for their ability to identify crucial TPS errors.
Methods: Over 50 IMRT fields, from low-modulation breast electronic compensation to high-modulation head/neck plans, were measured and analyzed via a commercial portal dose prediction package and a commercial portal dose conversion package. Various beam characteristics and dose calibration errors were induced into the TPS, then dose planes and dose predictions for several IMRT fields were recalculated for comparison to respective measurements. A common diode array system was also used in these studies, representing a standard for dose comparison.
Results: For IMRT fields with no induced errors, both EPID dosimetry systems perform similarly, except for large and/or off-axis fields where the EPID prediction system underestimates EPID response by up to 15%. In these cases, a 2D matrix correction developed by the authors and applied to individual EPID predictions brings the portal dose prediction system into good agreement with measurements. The portal dose prediction system, which uses actual fluence as an input, is able to catch some TPS errors, such as erroneous MLC transmission or dosimetric-leaf-gap values, but not errors downstream from the leaf-motion calculation. The portal dose conversion system, which compares measured absolute dose planes to TPS dose planes, is sensitive to errors anywhere in the fluence/dose calculation.
Conclusions: All EPID dosimetry systems are not created equal. Empirically-derived prediction algorithms may need substantial correction techniques, and may be insensitive to some TPS errors. Portal image conversion to dose matrix is a more comprehensive form of pretreatment IMRT QA.