Application of an inverse kernel concept to Monte Carlo based IMRT

Area of Science:

• Medical Physics
• Radiation Oncology
• Computational Imaging

Background:

• Pencil beam algorithms in inverse treatment planning can produce dose calculation errors, particularly where secondary electron equilibrium is absent.
• Monte Carlo (MC) simulations provide high accuracy but are computationally intensive, limiting their clinical use.

Purpose of the Study:

• To introduce and validate a new inverse kernel concept for precise and efficient inverse treatment planning.
• To compare the accuracy of the proposed inverse MC system against conventional pencil beam and collapsed-cone algorithms.

Main Methods:

• Developed an inverse kernel concept using precalculated MC simulations stored as binary trees.
• Implemented a workflow involving MC simulation, iterative optimization, segmentation, and reoptimization (Inverse Kernel Optimization - IKO).
• Applied the IKO method to a lung cancer case for evaluation.

Main Results:

• The IKO method demonstrated MC precision with acceptable computation times and memory usage.
• Comparisons revealed significant differences and systematic errors in plans generated by pencil beam and collapsed-cone algorithms versus IKO.
• Dose-volume and dose-difference histograms confirmed the superiority of the inverse MC approach.

Conclusions:

• Inverse Kernel Optimization (IKO) presents a precise, non-hybrid, inverse MC treatment planning system.
• The IKO system addresses current clinical needs by enabling multiple optimization steps after a single MC simulation.
• This method offers a viable alternative to existing algorithms, improving dose calculation accuracy in complex scenarios.