Room: AAPM ePoster Library
Purpose: To perform an in-depth comparison of three different Artificial Intelligence based decision support tools for head and neck treatment planning implemented in our clinic.
Methods: Models 1 and 2 are commercial products and model 3 in an in-house developed software. Model 1 extracts the features and objectives based on relationships between features and objectives learned from historical data. Model 2 utilizes machine-learning to define the planning objectives by matching the new patient to historical patients. Model 3 is an in house deep-learning dose prediction algorithm. 300 clinical plans were used for training and twenty patients were used for testing and evaluating the three models. Dose-volume constraints of high-impact OARs predicted by the three models were compared with the delivered clinical plans.
Results: Models 1 and 3 produce plans representing the average of the plans used to train the model. Model 2 matches more closely to the clinical intention of the physician by selecting the nearest treated patient. Models 2 and 3 are more flexible in accepting a wide range of plans with different clinical goals while model 1 requires the plans to be carefully selected for more specific prescription dose schemes. The comparison of predicted DVH values are presented. Model 1 is a planner oriented workflow which is advantageous to standardize the plan quality. However, the planner’s planning approach is restricted. Model 2 is physician oriented workflow, which has less impact in planner’s workflow and is well accepted and easier to implement. Our in-house model combines the advantageous of the two.
Conclusion: Decision support tools based on AI are valuable additions to the clinic that can increase the speed, consistency and quality of clinical head and neck plans. Each tool had different pros and cons but all created plans near a clinically acceptable plan.
TH- External Beam- Photons: Treatment planning using machine learning/Knowledge Based Planning/automation