A two-stage sequential linear programming approach to IMRT dose optimization.

Area of Science:

• Medical Physics
• Radiation Oncology
• Computational Biology

Background:

• Conventional intensity-modulated radiation therapy (IMRT) planning uses a two-stage process with approximate calculations, potentially leading to discrepancies in dose distribution.
• Heuristic optimization methods are often employed due to computational expense, but they may not explore the full solution space.

Purpose of the Study:

• To modify the traditional two-stage IMRT optimization process by integrating accurate Monte Carlo dose calculations and a sequential linear programming (SLP) approach.
• To evaluate the performance of the enhanced SLP method against conventional IMRT planning, direct-aperture, and segment weight optimization approaches.

Main Methods:

• Augmented the second stage of IMRT planning with Monte Carlo-based kernel-superposition dose calculations for physical beam apertures.
• Employed an exact mathematical programming-based sequential optimization approach using linear programming (SLP).
• Tested the approach on three clinical cases with multileaf collimator constraints from two vendors.

Main Results:

• The SLP approach demonstrated superior critical structure sparing compared to conventional IMRT, direct-aperture, and segment weight optimization methods across all tested cases.
• Convergence of the SLP approach was successfully demonstrated.
• The developed SLP method has been integrated into a commercial treatment planning system.

Conclusions:

• The enhanced two-stage IMRT optimization using SLP offers improved dose distribution and critical structure sparing.
• The SLP approach provides a more accurate and effective method for clinical IMRT planning.
• Integration into commercial systems suggests clinical viability and potential for widespread adoption.