Algorithms for the optimization of RBE-weighted dose in particle therapy

Area of Science:

• Medical Physics
• Computational Biology
• Radiotherapy

Background:

• Nonlinear optimization of dose is crucial for particle therapy.
• Accurate calculation of biological effects, like those from carbon ions using the Local Effect Model, necessitates iterative methods.
• Existing treatment planning systems require efficient optimization algorithms.

Purpose of the Study:

• To evaluate and implement advanced algorithms for nonlinear optimization of RBE-weighted dose in particle therapy.
• To compare the computational performance of various optimization algorithms within the TRiP98 treatment planning system.
• To enhance the speed and efficiency of dose optimization for complex treatment plans.

Main Methods:

• Implementation of BFGS and conjugate gradient algorithms into GSI's TRiP98 treatment planning system.
• Modification of standard iterative procedures to enhance convergence speed.
• Performance evaluation based on iteration count and computation time for carbon ion therapy dose calculations using the Local Effect Model.

Main Results:

• The Fletcher-Reeves variant of the method of conjugated gradients demonstrated superior computational performance.
• Computation times were reduced by a factor of 4 compared to the previously used method of steepest descent.
• Optimized complex treatment plans can now be achieved within minutes, yielding favorable dose distributions.

Conclusions:

• The Fletcher-Reeves algorithm offers a significant speed-up for RBE-weighted dose optimization in particle therapy.
• Faster optimization solvers are essential for addressing future challenges in particle therapy dose optimization.
• The implemented methods allow for efficient optimization of complex treatment plans, improving clinical workflow.