Evaluation of an artificial intelligence guided inverse planning system: clinical case study

Area of Science:

• Medical Physics
• Radiation Oncology
• Artificial Intelligence

Background:

• Inverse treatment planning in radiation therapy requires complex parameter adjustments.
• Manual parameter optimization is time-consuming and relies heavily on user expertise.
• Automating this process can potentially improve efficiency and consistency.

Purpose of the Study:

• To implement and evaluate an AI-guided method for automated parameter adjustment in inverse treatment planning.
• To compare the dosimetric outcomes of AI-automated plans versus manually generated plans.
• To assess the feasibility of AI in optimizing radiation therapy treatment plans.

Main Methods:

• An AI-driven fuzzy inference system was developed to adjust inverse planning parameters through iterative loops.
• Physician-defined dose constraints for the planning target volume (PTV) and organs at risk (OARs) guided the optimization process.
• Four clinical cases were evaluated, comparing automated plans with manually optimized plans.

Main Results:

• AI-automated plans showed a dose increase of up to 10% for 95% of the PTV compared to manual plans.
• Averaged dose reduction was observed for critical organs, though not universally improved due to current AI limitations.
• No significant difference in plan dose was found for normal tissues between automated and manual methods.

Conclusions:

• The AI-guided method successfully automated basic parameter adjustment in inverse planning, yielding comparable plan doses to manual methods.
• Further development, including case-specific inference rules, is necessary for full automation of the inverse planning process.
• AI shows promise for enhancing efficiency and consistency in radiation therapy planning.