Noninferiority Study of Automated Knowledge-Based Planning Versus Human-Driven Optimization Across Multiple Disease...

Area of Science:

• Radiation Oncology
• Medical Physics
• Cancer Treatment Planning

Background:

• Automated knowledge-based planning (KBP) aims to improve efficiency and quality in radiation therapy.
• Human-driven planning is the traditional method for developing radiation treatment plans.

Purpose of the Study:

• To assess if automated knowledge-based planning (KBP) is noninferior to human-driven planning for multiple cancer sites.
• To determine if KBP systematically improves dosimetric plan quality and reduces variability.

Main Methods:

• Developed clinical KBP automated planning for prostate, prostatic fossa, lung, and head and neck cancers.
• Conducted blinded plan selection by physicians comparing KBP and human-generated plans (145 cases).
• Analyzed plan quality and variability using dosimetric metrics pre- and post-KBP implementation.

Main Results:

• KBP achieved a noninferior success rate (57.2%) compared to human planning, with superiority in head and neck and lung SBRT.
• Significant differences observed in organ-at-risk sparing (KBP) and target homogeneity (human plans).
• Post-KBP implementation showed reduced organ doses and quality metric variability for prostate and head and neck.

Conclusions:

• Fully automated KBP is noninferior to human-driven plan optimization across multiple disease sites.
• KBP introduction systematically enhances treatment plan quality and reduces variability in radiation therapy.