Statistical evaluation of worst-case robust optimization intensity-modulated proton therapy plans using an exhaustive sampling approach

Zhiyong Yang, Heng Li, Yupeng Li, Yuting Li, Yu Chang, Qin Li, Kunyu Yang, Gang Wu, Narayan Sahoo, Falk Poenisch, Michael Gillin, X. Ronald Zhu, Xiaodong Zhang

Research output: Contribution to journalArticle

Abstract

Purpose: To assess the worst-case robust optimization IMPT plans with setup and range uncertainties and to test the hypothesis that the worst-case robust optimization strategies could cover most possible setup and range uncertainties in the real scenarios. Methods: We analyzed the nominal and worst-case robust optimization IMPT plans of seven patients with head and neck cancer patients. To take uncertainties into account for the dose calculation, we performed a comprehensive simulation in which the dose was recalculated 625 times per given plan using Gaussian systematic setup and proton range uncertainties. Subsequently, based on the simulation results, we calculated the target coverage in all perturbation scenarios, as well as the ratios of target coverage located within the threshold of eight worst-case scenarios. We set the criteria for the optimized plan to be the ratios of 1) the dose delivered to 95% (D95%) of clinical target volumes 1 and 2 (CTV1 and CTV2) above 95% of the prescribed dose, and 2) the D95% of clinical target volume 3 (CTV3) above 90% of the prescribed dose in worst-case situations. Results: The probability that the perturbed-dose indices of the CTVs in each scenario were within the worst-case scenario limits ranged from 89.51 to 91.22% for both the nominal and worst-case robust optimization IMPT plans. A quartile analysis showed that the selective robust optimization IMPT plans all had higher D95% values for CTV1, CTV2, and CTV3 than did the nominal IMPT plans. Conclusions: The worst-case strategy for robust optimization is adequately models and covers most of the setup and range uncertainties for the IMPT treatment of head and neck patients in our center.

Original languageEnglish (US)
Article number129
JournalRadiation Oncology
Volume14
Issue number1
DOIs
StatePublished - Jul 19 2019
Externally publishedYes

Fingerprint

Proton Therapy
Uncertainty
Head and Neck Neoplasms
Protons
Neck
Head

Keywords

  • Intensity-modulated proton therapy
  • Proton dose uncertainties
  • Robust optimization
  • Worst-case scenarios

ASJC Scopus subject areas

  • Oncology
  • Radiology Nuclear Medicine and imaging

Cite this

Statistical evaluation of worst-case robust optimization intensity-modulated proton therapy plans using an exhaustive sampling approach. / Yang, Zhiyong; Li, Heng; Li, Yupeng; Li, Yuting; Chang, Yu; Li, Qin; Yang, Kunyu; Wu, Gang; Sahoo, Narayan; Poenisch, Falk; Gillin, Michael; Zhu, X. Ronald; Zhang, Xiaodong.

In: Radiation Oncology, Vol. 14, No. 1, 129, 19.07.2019.

Research output: Contribution to journalArticle

Yang, Z, Li, H, Li, Y, Li, Y, Chang, Y, Li, Q, Yang, K, Wu, G, Sahoo, N, Poenisch, F, Gillin, M, Zhu, XR & Zhang, X 2019, 'Statistical evaluation of worst-case robust optimization intensity-modulated proton therapy plans using an exhaustive sampling approach', Radiation Oncology, vol. 14, no. 1, 129. https://doi.org/10.1186/s13014-019-1335-8
Yang, Zhiyong ; Li, Heng ; Li, Yupeng ; Li, Yuting ; Chang, Yu ; Li, Qin ; Yang, Kunyu ; Wu, Gang ; Sahoo, Narayan ; Poenisch, Falk ; Gillin, Michael ; Zhu, X. Ronald ; Zhang, Xiaodong. / Statistical evaluation of worst-case robust optimization intensity-modulated proton therapy plans using an exhaustive sampling approach. In: Radiation Oncology. 2019 ; Vol. 14, No. 1.
@article{db134b9f2d8f422688b28c333fa3b3bb,
title = "Statistical evaluation of worst-case robust optimization intensity-modulated proton therapy plans using an exhaustive sampling approach",
abstract = "Purpose: To assess the worst-case robust optimization IMPT plans with setup and range uncertainties and to test the hypothesis that the worst-case robust optimization strategies could cover most possible setup and range uncertainties in the real scenarios. Methods: We analyzed the nominal and worst-case robust optimization IMPT plans of seven patients with head and neck cancer patients. To take uncertainties into account for the dose calculation, we performed a comprehensive simulation in which the dose was recalculated 625 times per given plan using Gaussian systematic setup and proton range uncertainties. Subsequently, based on the simulation results, we calculated the target coverage in all perturbation scenarios, as well as the ratios of target coverage located within the threshold of eight worst-case scenarios. We set the criteria for the optimized plan to be the ratios of 1) the dose delivered to 95{\%} (D95{\%}) of clinical target volumes 1 and 2 (CTV1 and CTV2) above 95{\%} of the prescribed dose, and 2) the D95{\%} of clinical target volume 3 (CTV3) above 90{\%} of the prescribed dose in worst-case situations. Results: The probability that the perturbed-dose indices of the CTVs in each scenario were within the worst-case scenario limits ranged from 89.51 to 91.22{\%} for both the nominal and worst-case robust optimization IMPT plans. A quartile analysis showed that the selective robust optimization IMPT plans all had higher D95{\%} values for CTV1, CTV2, and CTV3 than did the nominal IMPT plans. Conclusions: The worst-case strategy for robust optimization is adequately models and covers most of the setup and range uncertainties for the IMPT treatment of head and neck patients in our center.",
keywords = "Intensity-modulated proton therapy, Proton dose uncertainties, Robust optimization, Worst-case scenarios",
author = "Zhiyong Yang and Heng Li and Yupeng Li and Yuting Li and Yu Chang and Qin Li and Kunyu Yang and Gang Wu and Narayan Sahoo and Falk Poenisch and Michael Gillin and Zhu, {X. Ronald} and Xiaodong Zhang",
year = "2019",
month = "7",
day = "19",
doi = "10.1186/s13014-019-1335-8",
language = "English (US)",
volume = "14",
journal = "Radiation Oncology",
issn = "1748-717X",
publisher = "BioMed Central",
number = "1",

}

TY - JOUR

T1 - Statistical evaluation of worst-case robust optimization intensity-modulated proton therapy plans using an exhaustive sampling approach

AU - Yang, Zhiyong

AU - Li, Heng

AU - Li, Yupeng

AU - Li, Yuting

AU - Chang, Yu

AU - Li, Qin

AU - Yang, Kunyu

AU - Wu, Gang

AU - Sahoo, Narayan

AU - Poenisch, Falk

AU - Gillin, Michael

AU - Zhu, X. Ronald

AU - Zhang, Xiaodong

PY - 2019/7/19

Y1 - 2019/7/19

N2 - Purpose: To assess the worst-case robust optimization IMPT plans with setup and range uncertainties and to test the hypothesis that the worst-case robust optimization strategies could cover most possible setup and range uncertainties in the real scenarios. Methods: We analyzed the nominal and worst-case robust optimization IMPT plans of seven patients with head and neck cancer patients. To take uncertainties into account for the dose calculation, we performed a comprehensive simulation in which the dose was recalculated 625 times per given plan using Gaussian systematic setup and proton range uncertainties. Subsequently, based on the simulation results, we calculated the target coverage in all perturbation scenarios, as well as the ratios of target coverage located within the threshold of eight worst-case scenarios. We set the criteria for the optimized plan to be the ratios of 1) the dose delivered to 95% (D95%) of clinical target volumes 1 and 2 (CTV1 and CTV2) above 95% of the prescribed dose, and 2) the D95% of clinical target volume 3 (CTV3) above 90% of the prescribed dose in worst-case situations. Results: The probability that the perturbed-dose indices of the CTVs in each scenario were within the worst-case scenario limits ranged from 89.51 to 91.22% for both the nominal and worst-case robust optimization IMPT plans. A quartile analysis showed that the selective robust optimization IMPT plans all had higher D95% values for CTV1, CTV2, and CTV3 than did the nominal IMPT plans. Conclusions: The worst-case strategy for robust optimization is adequately models and covers most of the setup and range uncertainties for the IMPT treatment of head and neck patients in our center.

AB - Purpose: To assess the worst-case robust optimization IMPT plans with setup and range uncertainties and to test the hypothesis that the worst-case robust optimization strategies could cover most possible setup and range uncertainties in the real scenarios. Methods: We analyzed the nominal and worst-case robust optimization IMPT plans of seven patients with head and neck cancer patients. To take uncertainties into account for the dose calculation, we performed a comprehensive simulation in which the dose was recalculated 625 times per given plan using Gaussian systematic setup and proton range uncertainties. Subsequently, based on the simulation results, we calculated the target coverage in all perturbation scenarios, as well as the ratios of target coverage located within the threshold of eight worst-case scenarios. We set the criteria for the optimized plan to be the ratios of 1) the dose delivered to 95% (D95%) of clinical target volumes 1 and 2 (CTV1 and CTV2) above 95% of the prescribed dose, and 2) the D95% of clinical target volume 3 (CTV3) above 90% of the prescribed dose in worst-case situations. Results: The probability that the perturbed-dose indices of the CTVs in each scenario were within the worst-case scenario limits ranged from 89.51 to 91.22% for both the nominal and worst-case robust optimization IMPT plans. A quartile analysis showed that the selective robust optimization IMPT plans all had higher D95% values for CTV1, CTV2, and CTV3 than did the nominal IMPT plans. Conclusions: The worst-case strategy for robust optimization is adequately models and covers most of the setup and range uncertainties for the IMPT treatment of head and neck patients in our center.

KW - Intensity-modulated proton therapy

KW - Proton dose uncertainties

KW - Robust optimization

KW - Worst-case scenarios

UR - http://www.scopus.com/inward/record.url?scp=85069531021&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85069531021&partnerID=8YFLogxK

U2 - 10.1186/s13014-019-1335-8

DO - 10.1186/s13014-019-1335-8

M3 - Article

C2 - 31324257

AN - SCOPUS:85069531021

VL - 14

JO - Radiation Oncology

JF - Radiation Oncology

SN - 1748-717X

IS - 1

M1 - 129

ER -