Non-coplanar automatic beam orientation selection in cranial IMRT: a practical methodology.

Area of Science:

• Medical Physics
• Radiation Oncology
• Computer Vision

Background:

• Optimizing beam orientations in cranial intensity-modulated radiation therapy (IMRT) is complex.
• Manual planning requires experienced radiation physicists and significant computation time.
• Non-coplanar beam arrangements offer potential advantages but increase planning complexity.

Purpose of the Study:

• To develop and evaluate an automatic beam selection (ABS) method for non-coplanar cranial IMRT.
• To achieve treatment plan quality comparable to that of expert physicists.
• To reduce computation time for beam orientation optimization.

Main Methods:

• A geometry-based methodology integrating computer vision, beam's eye view techniques, and neural networks.
• Development of an automatic beam selection (ABS) algorithm for non-coplanar beam orientations.
• Extensive testing on 14 patients with varying complexity, comparing non-coplanar vs. coplanar plans.

Main Results:

• The ABS process generated cranial IMRT plans comparable in quality to those created by experienced physicists.
• Optimization was achieved in clinically useful computation times, typically under 1 minute.
• The method consistently yielded good clinical results, demonstrating its robustness across diverse patient cases.
• Comparisons with coplanar plans identified specific scenarios where simpler coplanar arrangements are advantageous.

Conclusions:

• The proposed ABS methodology is effective for automatic beam orientation selection in cranial IMRT.
• The system produces high-quality treatment plans efficiently, making it suitable for clinical application.
• The study provides insights into the trade-offs between coplanar and non-coplanar plans for cranial lesions.