Room: Exhibit Hall | Forum 4
Purpose: Advanced radiotherapy techniques such as IMRT and VMAT necessitate patient-specific quality assurance (QA) tests. The most commonly used method of evaluating patient-specific QA is gamma-index analysis. Despite its widespread use, gamma-index analysis has been shown to have low sensitivity to errors outside of the therapeutic dose region. The objective of this study was to generalize gamma-index analysis to work simultaneously in all (i.e., high- and low-dose) regions of dose distributions spanning three or more orders of magnitude.
Methods: To do this, we extended the gamma index to include an additional criterion specifically for the low-dose region (i.e., D â‰¤ 5%). We also proposed methods to objectively select the appropriate magnitudes for each criterion. We demonstrated our new extended-gamma-index methods with two steps. First, we found the appropriate extended-gamma-index criteria for comparing measured dose profiles to those calculated using a physics-based analytical model at a range of selected passing rates. Second, we used these criteria to evaluate model-calculated doses when compared to an independently measured dose profile. Finally, we investigated the extended-gamma-index passing-rate sensitivity to dose-threshold selection.
Results: The objective search for appropriate agreement criteria revealed that criteria were sensitive to the selected passing rate. Further, the passing rate was sensitive to errors in the low-dose region of the independently measured dose profile. Additionally, extended-gamma-index passing rates were not sensitive to dose-threshold selection in the range of 5-50%.
Conclusion: The major finding of this work was that the new extended-gamma-index analysis was able to evaluate dosimetric agreement in both the high- and low-dose regions of a dose profile simultaneously. These methods could find utility not only in clinical quality assurance, but also in other related fields for which out-of-field dose magnitudes are of importance and different levels of agreement may be appropriate, including epidemiological studies or algorithm development and implementation.