GPU-accelerated FREDopt package for simultaneous dose and LET_dproton radiotherapy plan optimization via...

Area of Science:

• Medical Physics
• Computational Biology
• Radiotherapy Research

Background:

• Intensity-modulated proton therapy (IMPT) planning requires optimizing both radiation dose and linear energy transfer (LETd).
• Simultaneous optimization of dose and LETd is challenging but crucial for minimizing normal tissue toxicity.
• Existing optimization methods can be computationally intensive and may not fully address LETd reduction.

Purpose of the Study:

• Introduce FREDopt, a novel GPU-accelerated open-source software for simultaneous proton dose and LETd optimization in IMPT.
• Develop and validate a superiorization-based feasibility-seeking algorithm for efficient treatment plan optimization.
• Evaluate the clinical applicability and computational performance of FREDopt.

Main Methods:

• FREDopt was developed in Python, utilizing CuPy for GPU acceleration and fast Monte Carlo simulations.
• A treatment plan optimization workflow involving pre-optimization and a novel superiorization of feasibility-seeking algorithms was implemented.
• The software was validated on clinical patient treatment plans, comparing dose and LETd distributions before and after reoptimization.

Main Results:

• Simultaneous dose and LETd optimization using FREDopt significantly reduced LETd and the product of dose × LETd in organs at risk.
• Target dose conformity was preserved during the optimization process.
• Computational times ranged from 14 to 50 minutes, demonstrating clinical feasibility.

Conclusions:

• FREDopt provides an efficient and effective solution for simultaneous dose and LETd optimization in IMPT.
• The novel superiorization approach enhances feasibility-seeking, offering an alternative to complex constrained optimization.
• The open-source nature of FREDopt facilitates further research and clinical implementation in proton therapy.