Radiographers' perceptions on the challenges and new technologies for organ at risk contouring

Abstract

Purpose: This study aimed to evaluate the challenges encountered by local radiographers while defining the organs at risk (OARs) and to explore the use of techniques that could be used to facilitate the process. Methodology A self-designed online questionnaire was used to collect the data for this study. The first part of the questionnaire evaluated the demographics and experiences of radiographers. In the second part of the questionnaire, the radiographers were asked to rate the difficulties encountered when contouring specific OARs using a Likert scale from 0 (Not Difficult) to 4 (Very Difficult). Finally, radiographers were asked to rate the impact of introducing specific tools (such as artificial intelligence), training, and workflows within the department on facilitating the definition of OARs. The data were summarised using means +/- standard deviation, and the impact of experience on the perceived difficulty in contouring the OARs was assessed using the Kruskal Wallis Test. Results: The parotid gland (2.57 +/- 0.73) was identified as the most difficult OARs to contour, followed by the stomach (2.30 +/- 1.02), brainstem (2.0 +/- 0.80), bowel bag (2.0 +/- 0.74), bladder (0.04 +/- 0.21), femoral head (0.04 +/- 0.21), and body outline (0.04 +/- 0.21). Radiographers with more than 5 years of experience found it easier to contour most OARs, especially the two most difficult structures. Radiographers strongly agreed that contouring manually is a demanding, subjective, and time-consuming task. In terms of AI software, training was considered too laborious, and contouring accuracy received poor results. The introduction of peer-review was deemed as having the highest impact on reducing contouring difficulty. This was followed by provision of courses in OAR contouring, training on the use of AI contouring software, introduction of internal audits to review contours, implementation of more detailed protocols, increased staff rotation, purchasing auto-contouring software, and comparing contours with other departments. Conclusion: The parotid was identified as the most difficult OARs to contour by radiographers. However, more experienced radiographers found it less difficult to contour OARs. Regular feedback, training, and implementation of new AI software could be used to facilitate the contouring process.