The implementation of knowledge-based planning with partial OAR contours for prostate radiotherapy

Area of Science:

• Radiation Oncology
• Medical Physics
• Image-guided Therapy

Background:

• Observer variability in contouring organs-at-risk (OARs) poses challenges in radiotherapy treatment planning.
• Partial contouring, focusing on uninvolved OAR regions, is proposed to mitigate contour uncertainty.

Purpose of the Study:

• To evaluate the efficacy of partial contours in radiotherapy treatment planning.
• To compare dosimetry and contour concordance between partial and full OAR contouring methods.

Main Methods:

• Developed Rapid Plan (RP) and Rapid Plan partial uninvolved (RP_Part_Un) models using knowledge-based planning (KBP) and artificial intelligence (AI) auto-segmentation.
• Retrospectively analyzed 60 prostate cancer patients using three plan types: RP, RP_Part_Un, and MIM AI_Part_Un.
• Compared OAR sparing using dose metrics (Dmax, Dmean, V10-70Gy) and evaluated contour agreement using Dice Similarity Coefficient (DSC) and Overlay Index (OI).

Main Results:

• Partial contours demonstrated significantly increased concordance (higher DSC and OI) compared to full contours.
• RP_Part_Un plans showed reduced bladder dose metrics in patients with SpaceOAR and reduced rectal dose metrics in patients without SpaceOAR.
• MIM AI_Part_Un plans achieved lower rectal maximum dose in both patient groups.

Conclusions:

• Partial contours, defined by their distance from the planning target volume (PTV), provide dosimetry comparable to fully contoured plans.
• This method shows potential for improving accuracy and efficiency in radiotherapy treatment planning.