A Systematic Analysis of IMRT QA Results

Abstract

A commercial 2-D diode array (Mapcheck, Sun Nuclear) has been used for IMRT plan QA at the University of Florida. By retrospectively reviewing the QA data we have systematically evaluated factors that influence quantitative differences between calculation and measurement. For each field, a planar dose was calculated using Philips Pinnacle treatment planning system (TPS) at 10 cm depth, 100 cm SAD in solid water with normal incidence of the gantry and Mapcheck measurement was performed as planned. Absolute dose comparison was made between the calculated and measured with a 3 % and 3 mm criteria. A threshold was set at 10 % of normalization point dose. The pass rates were categorized into the versions of the TPS and IMRT delivery system. We also analyzed the outliers to see if we could a priori predict the differences between measured and calculated using a recently introduced dose uncertainty model (Medical Physics 2005, pp 1747). A total of 417 plans with 2210 beams were analyzed; 57.4 % of all beams had pass rates of at least 95 %; 36.2 % were between 85 % and 95 %; 6.4 % were below 85 %. The pass rate correlated strongly with the accuracy of beam modeling in TPS. With explicit modeling of MLC (e.g., MLC leaf-end and leaf-side effects), the TPS showed a better pass rate (60.3 % vs. 52.7 % in percent of having more than 95 % pass rate). The failure rate increased with the complexity of intensity-modulation. It also increased with the larger uncertainty of MLC beam model. In conclusion, the accuracy of IMRT delivery is strongly correlated to the accuracy of MLC beam model in TPS. The pass rate for patient specific QA is strongly influenced by the complexity of the intensity modulation in a field.