Click here to


Are you sure ?

Yes, do it No, cancel

Validation of An AI-Driven Treatment Planning System for Adaptive Radiotherapy

E Pryser*, B Cai, F Reynoso, E Laugeman, L Henke, H Kim, S Mutic, G Hugo, Washington University School of Medicine, St. Louis, MO


(Sunday, 7/12/2020)   [Eastern Time (GMT-4)]

Room: AAPM ePoster Library

Purpose: work describes commissioning and validation tests of the treatment planning component of the Ethos Adaptive Therapy system.

Methods: Validation tests of the Ethos Treatment Planning System (TPS) were designed to: (a) evaluate dosimetric accuracy and (b) confirm that the features of the software operate in a safe, consistent, and expected manner. Dosimetric accuracy was evaluated using measurements consistent with MPPG5.a recommendations and performed with a microDiamond detector, along with comparisons of dose calculated by Ethos and by an external TPS (Eclipse) for both IMRT and VMAT techniques to various treatment sites. System behavior tests included basic contouring on a known-volume phantom, contour integrity after importing or exporting from an external software, generation of contours with margin expansions, CT density assignment by material, identification and assignment of simulation isocenter, and DVH metrics reporting.

Results: On average, the difference in measured and predicted dose for the MPPG5.a fields was 1.3%, with the greatest discrepancy belonging to a circular field with diameter 2cm (?=2.83%). Dose calculated by the Ethos TPS was consistent with the Eclipse calculated dose, with an average point dose difference of 1.2%. The difference between the isocenter position assigned by Ethos and the expected isocenter position averaged 0.1cm. While the volume of structures generated using symmetric margins or small asymmetric margins (=1cm) showed acceptable agreement between the two planning systems, there were differences (?(V,Eclipse,Ethos)=13.5%) in the structures generated when asymmetric margins >1cm were applied. Dose metrics reported by Ethos agreed with those reported by Eclipse, giving an average difference in reported metric of ?(DVH)=1.25%.

Conclusion: Validation tests show that the treatment planning feature of the Ethos system calculates dose accurately. While there are some differences in contour generation, tests confirmed that the system operates in a manner that is consistent and safe for clinical use.

Funding Support, Disclosures, and Conflict of Interest: GH: Honoraria and travel costs, Varian Medical Systems; Research grants: Varian Medical Systems, Viewray, Siemens


Commissioning, Treatment Planning, Validation


TH- External Beam- Photons: Development (new technology and techniques)

Contact Email