Oral Presentation 37th TROG Cancer Research Annual Scientific Meeting 2025

Implementation of an automated contour quality assurance tool in the TROG 18.01 NINJA trial (#8)

Lois Holloway 1 2 3 4 , Phillip Chlap 1 2 3 , Hang Min 1 2 5 , Mark Sidhom 2 3 , Jarad Martin 6 7 , Angela Whitehead 8 , Alisha Moore 8 , Annette Haworth 2 4 , Martin A Ebert 9 10 11 , Shalini K Vinod 1 2 3 , Jason Dowling 2 5
  1. Ingham Institute of Applied Medical Research, Liverpool, NSW, Australia
  2. University of New South Wales, Sydney, NSW, Australia
  3. Liverpool and Macarthur Cancer Therapy Centre, Liverpool, NSW, Australia
  4. Institute for Medical Physics, School of Physics, The University of Sydney, Sydney, NSW, Australia
  5. Australian e-Health Research Centre, CSIRO, Brisbane, QLD, Australia
  6. Calvary Mater Newcastle Hospital, Newcastle, NSW, Australia
  7. School of Medicine and Public Health, University of Newcastle, Newcastle, NSW, Australia
  8. Trans-Tasman Radiation Oncology Group, Newcastle, NSW, Australia
  9. Sir Charles Gairdner Hospital, Perth, WA, Australia
  10. University of Western Australia, Perth, WA, Australia
  11. University of Wisconsin, Madison, WI, USA

Background
Contour quality assurance (CQA) is an essential component of modern radiotherapy clinical trials as it has the potential to influence patient and trial outcomes. Automation is desirable to overcome time and resource constraints of manual CQA. An automated approach for CQA of the Clinical Target Volume (CTV) on an MRI image was previously developed and implemented within the TROG 18.01 NINJA trial.

Aims
In this work we assess the feasibility and impact of the use of an automated CQA tool within NINJA.

Methods
The automated CQA tool was integrated within MIM Software (Cleveland, OH) at TROG. For trial cases which are sampled for review, DICOM data submitted by the treating centre is loaded into MIM and prepared by a TROG team member. This includes registering the CT image with the MRI if required. The MIM workflow which runs the automated tool is triggered and returns a PDF report containing the automated QA result, including whether the case passes or requires additional review. These reports were forwarded to the reviewing radiation oncologist to accompany their manual review.

Results
To date, the automated QA tool has been applied to 51 cases within the NINJA trial. The tool deemed 20 as acceptable with 31 requiring additional review. 2 cases flagged were determined to require a resubmission after manual review. While this indicates a relatively high false positive rate, these were quick to assess by a reviewer, and there were no false negatives. End user reviews found an efficient streamlining in the contour assessment processes.

Conclusions
The use of an automated CQA tool within the NINJA trial was demonstrated to be feasible. This tool has potential to remove the requirement for manual review for approximately 40% (20/51) of cases without missing a true contour violation, and speeds the assessment of possible violations.