Session Item

Automated treatment planning is not a new phenomenon in brachytherapy. It helps to calculate individual patient plans in order to maximize tumor control and minimize toxicity. Recent publications show that automated planning is finally turning from a research niche to real clinical practice.

Various research has shown that automated algorithms can calculated better plans, in terms of tumor coverage and sparing of healthy tissue, in a shorter time frame. Retrospective evaluation studies have shown their value for the acceptance of plans obtained by an automated algorithm, and the transition to clinical practise could be made. However, the idea prevails that automated planning, including artificial intelligence algorithms, are challenging to implement in clinical practice and might met resistance.
In March 2020, we clinically introduced ‘BRachytherapy via artificially Intelligent GOMEA-Heuristic based Treatment planning’ (BRIGHT) for prostate cancer patients in our institution.
Last two years we gained clinical experience and learned that when automated treatment planning is used in clinical practice new challenges arise. In example, the physician makes use of patient-specific information, experience and general knowledge that was not, or cannot be, captured by dose-volume criteria implemented in the new algorithm. The result was that the automated plans were manually fine-tuned when deemed necessary, which reduced the time savings. Furthermore, staffing, training and integration in an commercial treatment planning system appeared of greater influence on the performance than expected.
Based on our experience, additional optimization aims need to be implemented to further improve direct clinical applicability of treatment plans and process efficiency. The evaluation of plan selection criteria and reasons for manual fine-tuning taught us which additional criteria are deemed of importance, such as dose-volume aims for gross-tumour-volume, contiguous high-dose sub-volumes (“hotspots”), and high-dose regions in close proximity to organs at risk.
One of the most important differences with respect to manual planning was that there is a shift in the use of time. With BRIGHT, or other types of automated treatment planning algorithms, time can be spent on carefully choosing the desired plan, instead of adjusting one plan iteratively without having any indication one has actually arrived at the best possible plan for that specific patient and what alternatives are.
The main focus of this talk will be on the challenges of introducing automated planning and practical examples of our first experiences in the clinic.
Additionally will be addressed how one can overcome these challenges and what role the RTT, doctors and medical physicists play in turning the challenges into opportunities.