Online

ESTRO 2020

Session Item

Physics track: Dose measurement and dose calculation
9319
Poster
Physics
00:00 - 00:00
Validating a commercial solution for calculation-based QA of AI-driven online adaptive radiotherapy
Patrik Sibolt, Denmark
PO-1415

Abstract

Validating a commercial solution for calculation-based QA of AI-driven online adaptive radiotherapy
Authors: Lina Andersson.(Herlev & Gentofte Hospital, Radiotherapy Research Unit- Department of Oncology, Herlev, Denmark), Ulf Bjelkengren.(Herlev & Gentofte Hospital, Radiotherapy Research Unit- Department of Oncology, Herlev, Denmark), Lucie Calmels.(Herlev & Gentofte Hospital, Radiotherapy Research Unit- Department of Oncology, Herlev, Denmark), Grichar Valdes Santurio.(Herlev & Gentofte Hospital, Radiotherapy Research Unit- Department of Oncology, Herlev, Denmark), Patrik Sibolt.(Herlev & Gentofte Hospital, Radiotherapy Research Unit- Department of Oncology, Herlev, Denmark), Maria Sjölin.(Herlev & Gentofte Hospital, Radiotherapy Research Unit- Department of Oncology, Herlev, Denmark), David Sjöström.(Herlev & Gentofte Hospital, Radiotherapy Research Unit- Department of Oncology, Herlev, Denmark)
Show Affiliations
Purpose or Objective

Quality assurance of dynamic treatment plans in radiotherapy has traditionally been relying on phantom based measurements. However, modern digital treatment delivery systems have today evolved into inherently being safe and reliable, and together with an expansion of machine specific QA a tendency to rely on only calculation-based plan-specific QA can be observed. Furthermore, with online adaptive radiotherapy (oART) there is no longer the possibility to within a treatment session conduct measurement-based quality control (QC) of treatment plans generated on the anatomy of the day. This study aims at providing commissioning data supporting the implementation of a commercial solution for calculation-based QA of online adaptive radiotherapy.

Material and Methods

A set of 48 automatically generated treatment plans (9- and 12-field IMRT, 2- and 3-arc VMAT) for bladder and rectum patients, in both cases including simultaneous treatment of pelvic lymph nodes, were created using the Ethos system for oART (Varian Medical systems, Palo Alto, CA). Plans were exported to the Mobius3D system (v. 2.2) for independent dose calculation and measured using the Delta4 phantom (ScandiDos AB, Uppsala, Sweden) as well as portal dosimetry. The Ethos system applies final dose calculation based on the AcurosXB algorithm, while Mobius3D calculations are based on a collapsed-cone algorithm. Two- and three-dimensional gamma evaluations were conducted according to standard protocols in our clinic and final cross-comparison between the three QA systems were conducted.

Results

The Mobius3D calculations demonstrate average gamma passing rates above 97% (3%/3mm, local gamma) for all cases, comparing well with corresponding results for Delta4 measurements (3%/2mm, local gamma) (Table 1). Results also indicate that portal dosimetry is insensitive and does not detect variations in plan complexity and treatment delivery, with a 100% passing rate (3%/2mm, local gamma) which is not corresponding to Delta4 measurements, currently acting as standard for plan-specific QA in our clinic. Furthermore, as the independent 3D dose calculation is conducted in the patient geometry it provides valuable information of the origin of potential dose differences, enabling further investigation in to the potential clinical impact of the QC results. Overall, the results demonstrate high deliverability of the automatically generated Ethos plans for oART.
Table 1

Conclusion

Commissioning data for validation of the commercially available Mobius3D solution for calculation-based QA of oART treatment plans has been provided. The system has been demonstrated to be comparable to our current standard of practice for measurement-based plan-specific QC, and furthermore to be a safe and valuable tool for oART.