Encrypted login | home

Program Information

Quantifying Normal Tissue Sparing with 4D Robust Optimization of Intensity Modulated Proton Therapy

M Newpower

M Newpower*, S Ge , R Mohan , UT MD Anderson Cancer Center, Houston, TX


SU-F-T-187 (Sunday, July 31, 2016) 3:00 PM - 6:00 PM Room: Exhibit Hall

Purpose:To report an approach to quantify the normal tissue sparing for 4D robustly-optimized versus PTV-optimized IMPT plans.

Methods:We generated two sets of 90 DVHs from a patient’s 10-phase 4D CT set; one by conventional PTV-based optimization done in the Eclipse treatment planning system, and the other by an in-house robust optimization algorithm. The 90 DVHs were created for the following scenarios in each of the ten phases of the 4DCT: ± 5mm shift along x, y, z; ± 3.5% range uncertainty and a nominal scenario. A Matlab function written by Gay and Niemierko was modified to calculate EUD for each DVH for the following structures: esophagus, heart, ipsilateral lung and spinal cord. An F-test determined whether or not the variances of each structure’s DVHs were statistically different. Then a t-test determined if the average EUDs for each optimization algorithm were statistically significantly different.

Results:T-test results showed each structure had a statistically significant difference in average EUD when comparing robust optimization versus PTV-based optimization. Under robust optimization all structures except the spinal cord received lower EUDs than PTV-based optimization. Using robust optimization the average EUDs decreased 1.45% for the esophagus, 1.54% for the heart and 5.45% for the ipsilateral lung. The average EUD to the spinal cord increased 24.86% but was still well below tolerance.

Conclusion:This work has helped quantify a qualitative relationship noted earlier in our work: that robust optimization leads to plans with greater normal tissue sparing compared to PTV-based optimization. Except in the case of the spinal cord all structures received a lower EUD under robust optimization and these results are statistically significant. While the average EUD to the spinal cord increased to 25.06 Gy under robust optimization it is still well under the TD50 value of 66.5 Gy from Emami et al.

Funding Support, Disclosures, and Conflict of Interest: Supported in part by the NCI U19 CA021239.

Contact Email: