Global optimization for spot-based treatment planning.

Area of Science:

• Medical Physics
• Radiation Oncology
• Computational Imaging

Background:

• Spot-based radiotherapy, including Gamma Knife (GK) and GammaPod (GP), offers precise radiation delivery to targets while sparing healthy tissues.
• Global optimization of spot positions, shapes, and intensities in 3D is computationally intractable with conventional methods.
• Current approaches use sequential heuristic methods, limiting treatment plan optimization.

Purpose of the Study:

• To develop a novel framework for global optimization in spot-based treatment planning.
• To overcome the computational complexity of optimizing numerous spots in 3D radiotherapy.
• To enable clinically feasible, globally optimized treatment plans for modalities like GK and GP.

Main Methods:

• The framework utilizes kernel decomposition (KD) for dose calculation, modeling spots as scaled, shift-invariant kernels.
• Fast Fourier Transform (FFT) is integrated for efficient objective and derivative evaluations during optimization.
• The approach accommodates all spot candidates, reducing temporal complexity from O(N^6) to O(N^3 log N) compared to beamlet methods.

Main Results:

• The KD model requires minimal pre-processing with few reference kernels and a scale map.
• FFT-based optimization for 512x512 simulations completed in ~2 seconds, 100x faster than beamlet methods.
• Clinical cases showed improved plan quality, including better conformity and reduced integral dose, with solutions found within a minute.

Conclusions:

• The scaled shift-invariance and FFT framework represent a new paradigm for spot-based radiotherapy planning.
• This approach significantly reduces spatial and temporal complexities in treatment planning.
• Global optimization for spot-based radiotherapy is now clinically feasible, enhancing treatment precision and efficiency.