The use of mixed-integer programming for inverse treatment planning with pre-defined field segments

Area of Science:

• Radiation Oncology
• Medical Physics
• Computational Biology

Background:

• Traditional beamlet-based inverse treatment planning can generate complex intensity patterns that are challenging to deliver accurately.
• Existing methods for intensity-modulated radiation therapy (IMRT) optimization face difficulties in practical implementation due to delivery complexity.

Purpose of the Study:

• To develop and evaluate a segmental inverse planning approach for radiation therapy dose optimization.
• To compare the efficacy of mixed-integer programming (MIP) with traditional methods for optimizing radiation treatment plans.

Main Methods:

• A novel approach was developed where intensity maps are controlled by pre-defined field segments for dose optimization.
• A pool of allowable delivery segments was defined using simple rules, and mixed-integer programming (MIP) was employed to optimize segment weights.
• The optimization problem incorporated real variables for segment weights and binary variables for target and critical structure voxels, and was compared against the Cimmino projection algorithm and traditional beamlet-based IMRT.

Main Results:

• Segmental inverse planning produced treatment plans comparable to traditional beamlet-based IMRT plans in complex oropharyngeal cancer cases.
• Mixed-integer programming (MIP) effectively imposed dose-volume constraints and identified optimal solutions for feasible problems.
• The segmental technique demonstrated advantages in simplified dosimetry, quality assurance, and treatment delivery.

Conclusions:

• The segmental inverse planning approach using MIP offers a viable alternative to traditional beamlet-based methods for IMRT.
• This method simplifies complex treatment planning, improves deliverability, and enhances quality assurance processes.
• The MIP-based segmental technique provides a robust mechanism for optimizing radiation therapy plans while adhering to dose constraints.