Fifty-seventh annual meeting of the American association of physicists in medicine
SU-E-J-178: Towards ART - Bootstrap Based Estimation of Cumulated Dose Distribution
A lot of effort is put into diminishing the CTV-PTV margin and providing Adaptive RT in order to spare normal structures adjacent to the PTV. The purpose of this work was to test the hypothesis that bootstrap method can be used to estimate cumulated dose distribution, representing the whole treatment and that the quality of this estimation is independent on the CTV-PTV margin.
For 25 prostate patients treated in our clinic daily set-up errors were recorded. For each patient treatment plans for three techniques were prepared: 3D-CRT, IMRT and VMAT. We prepared plans for different CTV-PTV margins, for asymmetric margin (0.4, 0.7, and 0.7 cm for lateral, cranial-caudal, and AP respectively) and than for several symmetric ones of 0.0, 0.2, 0.4, 0.6, 0.8, 1.0 cm. For each treatment plan and each fraction the dose distribution was calculated with the isocenter corrected for the set-up error. Cumulated Dose Distribution (CDDall) was calculated as the sum of all dose distributions for each single fraction. Estimated Dose Distribution (EDDn) was calculated with bootstrap methodology. The EDDn and CDDall were compared using the 3D gamma concept (2mm and 2%) for each pair of dose distributions. The Result was treated as acceptable if at least 95% of voxels with gamma index ≤ 1 were obtained.
For 3D and IMRT EDDn calculated from at least 8 (3D) and 7 (IMRT) first fractions (EDDn>7 and EDDn>6), for all patients, but one, acceptable results were obtained. For VMAT our requirement was fulfilled for all patients and EDDn>2. Statistical tests show no significant difference between gamma results for different CTV-PTV margins.
The cumulated dose distribution CDDall can be estimated from at least 8 fractions regardless of the technique and CTV-PTV margin. For VMAT technique the cumulated dose distribution may be estimated with 3 fractions.