A Hamiltonian Engine for Radiotherapy Optimization

Area of Science:

• Medical Physics
• Computational Optimization
• Radiation Oncology

Background:

• Intensity-Modulated Radiation Therapy (IMRT) planning involves complex optimization problems.
• Existing Quantum Annealer (QA) approaches for IMRT optimization are limited by variable count and long computation times.
• Conventional hardware solutions are needed to overcome QA limitations in radiotherapy optimization.

Purpose of the Study:

• To introduce and evaluate the Hamiltonian Engine for Radiotherapy Optimization (HERO) platform for IMRT treatment planning.
• To compare HERO's performance against commercial software in terms of speed and plan quality.
• To explore HERO's potential for other radiotherapy techniques like Volumetric-Modulated Arc Therapy (VMAT).

Main Methods:

• HERO decomposes large binary optimization problems into sub-problems, approximating them with a quadratic pseudo-boolean function.
• The optimization is solved as a Quadratic Unconstrained Binary Optimization (QUBO) problem.
• HERO utilizes conventional hardware, operating at room temperature, to solve QUBO problems.

Main Results:

• HERO solved seven prostate IMRT cases (approx. 6000 beamlets each) in about 30 seconds per case.
• HERO achieved significantly lower objective function scores compared to the commercial software Eclipse.
• The platform demonstrated a substantial improvement in time-to-solution.

Conclusions:

• HERO is a promising hardware and software platform for radiation therapy optimization, specifically for IMRT.
• HERO offers a faster and potentially more effective alternative to existing optimization methods.
• The platform's capabilities may extend to VMAT and other complex radiotherapy planning scenarios.