An audit of the impact of the introduction of a commercial artificial intelligence driven auto-contouring tool into a radiotherapy department.

Abstract

Objectives: To audit prospectively the accuracy, time saving and utility of a commercial artificial intelligence auto-contouring tool (AIAC). To assess the reallocation of time released by AIAC.

Methods: We audited the perceived usefulness (PU), clinical acceptability and reallocation of time during the introduction of a commercial AIAC. The time from CT to plan completion (PPTT) was audited for several pathways.

Results: 248 patients and 32 staff were included in this audit. PU increased with exposure to AIAC (p < 0.05). For 80% of sites AIAC was timesaving and AI contours were clinically acceptable after minor edits. Edits had little impact on doses for the majority of cases. Median PPTT reduced by 5.5 (breast) and 9 (prostate) working days (p < 0.01). Radiographers spent more time on other tasks within planning. Oncologists improved their work-life balance and increased time spent on professional development and research by up to 2 hours per week.

Conclusions: All users of AIAC found it a useful tool and it improved their productivity. The contours were high quality and needed little editing. It reduced contouring time and reduced PPTT by several days in some cases. The reallocated time was staff group dependent.

Advances in knowledge: The time released by the use of AIAC can lead to a reduction in the PPTT by up to 9 days. It also improves the work-life balance of oncologists by reducing the time spent out of hours contouring.