Program Information
An Update On the Dosimetric Accuracy of Different Treatment Planning System Algorithms in the IROC Lung Phantom
S Kry*, A Molineu , P Alvarez , D Followill , UT MD Anderson Cancer Center, Houston, TX
Presentations
TH-CD-205-1 (Thursday, August 3, 2017) 10:00 AM - 12:00 PM Room: 205
Purpose: We previously reported on differences between treatment planning system algorithms in terms of calculation accuracy to the center of the target in the IROC lung phantom. Monte Carlo algorithms were found to be significantly more accurate than S/C and AAA algorithms, which systematically overestimated the dose by 3.7%. This work updates these findings based on several hundred more phantom irradiations, allowing different Monte Carlo algorithms to now be considered separately, and including results from the new Acuros dose calculation algorithm.
Methods: The expanded lung phantom cohort included 929 irradiations using AAA (427), S/C (360), Monte Carlo (89) or Acuros (63). The dose to the center of a cylindrical target (3 cm diameter and 5 cm long) was measured with two TLD. These measurements were compared with the institution’s dose predicted at the same location.
Results: AAA and S/C continued to overestimate the dose to the center of the target by 3.8% and 3.2% respectively. Different S/C algorithms showed some variations, with XiO providing the poorest agreement (3.9% overestimation; n=39) and Tomotherapy providing the best agreement (2.3% overestimation; n=53). This difference was statistically significant. Acuros and Monte Carlo showed statistically significantly better accuracy than AAA or S/C, but sub-group analysis found complexities to this generalization. Acuros showed average agreement within 0.9%. Monte Carlo results, on average, overestimated the dose by 1.8%, but this varied by specific algorithm. Multiplan agreed within 0.8% of measurement, but iPlan and Monaco overestimated the dose by 3% and 2.9% respectively.
Conclusion: Different algorithms show different levels of accuracy. This expanded cohort has reinforced the surprising limitations of S/C and AAA algorithms, revealed better agreement with the Acuros algorithm, and revealed inconsistencies between different Monte Carlo algorithms. Medical physicists need to continue to improve the accuracy of dose calculation algorithms used clinically.
Funding Support, Disclosures, and Conflict of Interest: This work was supported by grant # CA180803
Contact Email: