Abstract
Multi-objective optimisation using BRIGHT has proven insightful and effective in prostate cancer brachytherapy treatment planning. This study presents a novel, custom-developed graphical user interface (GUI) for BRIGHT applied to cervical cancer. The GUI enables plan navigation, pairwise comparisons, and dose distribution visualisation. End-user validation with a multidisciplinary team for ten patients showed an 'excellent' System Usability Scale (SUS) score of 83.3. BRIGHT plans were preferred over clinical plans in 80% of cases, demonstrating its clinical viability and user-friendliness.
Introduction
Brachytherapy (BT) is essential for treating cervical cancer, requiring a delicate balance between maximizing tumor radiation dose (coverage) and minimizing exposure to organs-at-risk (sparing), such as the bladder and rectum. Traditional planning is a manual, iterative process. While automated tools exist, they often lack the flexibility needed for clinical trade-offs. The BRIGHT (BRachytherapy via artificially Intelligent GOMEA-Heuristic based Treatment planning) system addresses this by generating a "Pareto front" of optimal plans, representing the best possible trade-offs between conflicting clinical objectives.
The BRIGHT Graphical User Interface
To make the complex mathematical optimization of BRIGHT clinically accessible, a custom GUI was developed. It features the "Golden Corner"—a 2D plot allowing clinicians to visualize the physical limits of achievable trade-offs for a specific patient. It also includes a "Difference View" for side-by-side plan comparisons using color-coded overlays and interactive re-optimization capabilities that allow users to refine plan sets in approximately 30 seconds.
Methods and Validation Study
The study involved a retrospective validation using data from 10 cervical cancer patients treated between 2023 and 2024. A multidisciplinary team consisting of radiation oncologists, medical physicists, and radiation therapy technologists evaluated the system by emulating clinical practice.
The evaluation focused on two main areas:
- Usability: Measured via the System Usability Scale (SUS) after users interacted with the interface.
- Clinical Preference: Conducted through blinded one-on-one comparisons between BRIGHT-generated plans and the original clinical plans created using standard manual optimization software.
Results
The GUI received a SUS score of 83.3, indicating "excellent" usability. Clinicians found the "Golden Corner" particularly valuable for quickly identifying whether planning aims were achievable or if concessions were necessary, preventing the "dead-end" sessions common in manual planning.
In blinded comparisons, the BRIGHT plan was preferred over the clinical plan in 8 out of 10 patients. In four of these cases, the differences were deemed clinically relevant. BRIGHT outperformed clinical practice in terms of the coverage-sparing trade-off in 50% of patients and performed equally well in the rest. Efficiency was also notably improved, with a median time of only 8.5 minutes spent in the GUI to finalize a plan.
Conclusion
BRIGHT, supported by its dedicated GUI, provides a user-friendly and clinically effective tool for cervical cancer brachytherapy planning. By allowing clinicians to navigate optimized trade-offs rather than manually iterating, the system improves both plan quality and workflow efficiency. Future work will focus on refining navigation transitions and extending the tool to other cancer types.
Summary updated