To Better Understand IMRT and Planned Dose Distributions: A Measure of Entropy
A Veres*, M Snyder, Wayne State University, Detroit, MISU-C-108-4 Sunday 1:00PM - 1:55PM Room: 108
Traditional IMRT QA procedures yield close approximations to the patients planned dose distribution. However, the relationship between QA data and the patients planned distribution is difficult to quantify. This study presents an approach to investigate the relationship between a planar QA distribution and the same doseplane within the patient using complexity analysis.
Shannon entropy was the decided metric to quantify the complexity of three calculated dose distributions: 1) the IMRT QA distribution, 2) the in-patient dose distributions with heterogeneity correction on, and 3) the in-patient dose distribution without heterogeneity correction. Varian Eclipse treatment planning system was used to construct the plans and calculate dose. MapCheck 1175 diode array was used for the IMRT QA acquisition. Global entropy was calculated without threshold for each distribution, and local entropies were found for low dose thresholds ranging from 0% to 100% of both prescribed and maximum dose. All calculations were done using MATLAB.
The QA data proved to have the highest global entropy of the three dose distributions analyzed. Local entropies, calculated using subsets of all dose points, converged at a low dose threshold of >~70% of maximum dose. For low dose thresholds <~70%, the entropies of both patient distributions remained similar, while the entropy for the QA distribution diverged, increasing in value.
The greater entropy values for the QA distribution at low dose thresholds <~70% suggests a more disordered dose distribution than the two planned dose distributions. This is thought to be due to the larger phantom size with respect to the patient, which tends to spread out the dose gradient outside the target. This result is suggestive that an optimal threshold for meaningful gamma analysis could be >70% of maximum dose, or where the QA and planned dose distributions entropies, or complexities, are accurately matched.