Physically constrained voxel-based penalty adaptation for ultra-fast IMRT planning

Area of Science:

• Medical Physics
• Radiation Oncology
• Computational Biology

Background:

• Conventional intensity-modulated radiation therapy (IMRT) planning is labor-intensive, relying on iterative manual optimization.
• Existing methods often struggle to efficiently incorporate physical constraints of photon irradiation into treatment planning.

Purpose of the Study:

• To develop and evaluate an automated algorithm for IMRT treatment planning.
• To specifically address the adaptation of voxel-based penalties in normal tissues, considering physical limitations.
• To assess the algorithm's performance in eradicating dose hot spots and its impact on target coverage and organs at risk.

Main Methods:

• Development of a physically constrained voxel-based penalty adaptation algorithm.
• Application of the algorithm to treatment planning studies involving 16 prostate and 5 head and neck cancer cases.
• Evaluation metrics included hot spot eradication, target coverage, target conformity, and dose-volume points for organs at risk.

Main Results:

• The algorithm successfully detected and eradicated hot spots in normal tissues.
• Target coverage and conformity were maintained.
• Negative effects on organs at risk were minimal, primarily affecting lower dose levels.
• Improved generation of class solutions for both prostate and head and neck indications was observed.
• Algorithm run-times were consistently under 20 seconds.

Conclusions:

• Physically constrained voxel-based penalty adaptation automates hot spot eradication in IMRT planning effectively.
• The method preserves or improves treatment plan quality while minimizing toxicity.
• The algorithm shows potential for reducing clinical workload and facilitating automated, adaptive radiation treatments.