Organ-at-risk contouring education in the era of AI-assisted practice: Insights from an Australian undergraduate radiation therapy program

Abstract

INTRODUCTION: The integration of artificial intelligence (AI) tools into radiation therapy workflows offers significant opportunities to improve efficiency by automating tasks such as contouring organs at risk (OARs). However, this also raises concerns regarding future practitioners' ability to critically evaluate auto-generated contours. This educational perspective examines how OAR contouring education can be integrated into undergraduate radiation therapy programs to support the development of foundational contouring competence alongside emerging technological capabilities. METHODS: An analysis of the OAR contouring curriculum at the authors' institution was conducted to assess the scope of contouring education and the technologies used in training, with consideration of contemporary practice. Students' preferred contouring methods, self-rated confidence in contouring different OARs, and factors influencing contour quality were also examined to identify educational considerations relevant to curriculum design. RESULTS: Analysis of the curriculum revealed the inclusion of a wide range of OARs across anatomical regions and imaging modalities, providing opportunities for progressive skill development and contouring experiences. Knowledge of anatomy was identified as the primary factor influencing contouring accuracy. Students expressed a preference for more efficient approaches, particularly guided contouring methods in which pre-generated contours could be edited. Analysis of students' confidence showed variability across OARs, with lower confidence reported for smaller and anatomically complex structures. CONCLUSIONS: The findings highlight the importance of foundational anatomy, progressive contouring experiences, and contour evaluation skills in undergraduate radiation therapy education. A scaffolded approach may help integrate these elements while preparing students for the critical use of AI contouring tools. PLAIN LANGUAGE SUMMARY: Artificial intelligence is increasingly being used in radiation therapy to help with tasks such as outlining normal body structures on medical images. This educational study reviewed an undergraduate training program to explore how students learn contouring, which is the process of marking important structures on scans for treatment planning. This study found that anatomy knowledge was the main factor affecting contour accuracy, students preferred being able to review and edit pre-generated contours, and confidence was lower for smaller or more complex structures. This matters because strong foundational skills and careful evaluation of artificial intelligence tools can help future practitioners deliver safe and accurate treatment.