Dose domain regularization of MLC leaf patterns for highly complex IMRT plans.

Area of Science:

• Radiation Oncology
• Medical Physics
• Computational Biology

Background:

• Automated intensity modulated radiation therapy (IMRT) planning allows for complex treatment strategies with numerous beams.
• Converting these complex fluence maps into deliverable multileaf collimator (MLC) segments presents a significant challenge.
• Simplifying MLC segments is crucial for improving delivery efficiency, accuracy, and overall plan quality.

Purpose of the Study:

• To introduce a novel method for regularizing fluence maps and simplifying MLC segments in IMRT.
• To maximize delivery efficiency, accuracy, and plan quality by addressing the fluence-to-segment conversion challenge.

Main Methods:

• Implemented a dose domain regularization approach minimizing segmentation-induced dose degradation.
• Utilized a total variation regularization to promote piecewise smoothness in fluence maps.
• Solved the optimization problem using the Chambolle-Pock algorithm, applied to patient cases with noncoplanar beams.

Main Results:

• The dose domain regularized (DDR) method improved average planning target volume dose homogeneity (D95/D5) from 0.814 to 0.878 across six patient cases.
• Regularized fluences demonstrated increased robustness to MLC sequencing, stratification, and small aperture removal.
• The DDR method consistently produced larger maximum and mean aperture sizes compared to direct MLC segmentation (DMS).

Conclusions:

• Formulated the fluence map to MLC segmentation as a dose domain optimization problem.
• Successfully transformed complex fluences into more robust maps using a primal-dual algorithm.
• Achieved fewer and larger MLC segments with minimal dose distribution degradation, enhancing IMRT delivery.