Room: Exhibit Hall | Forum 7
Purpose: This work introduces a new optimized approach for spot placement in pencil beam scanning (PBS) proton therapy that achieves the boundary conformity of a contoured approach and uses a fast-iterative method to optimally distribute the interior spots in a uniform fashion improving lateral dose falloff and sparing of nearby Organs at Risk (OARs).
Methods: This work breaks away from traditional grid-based spot placements and introduces a process to create optimized distributions of spots within a target for PBS proton therapy. It was developed using Delaunay Triangulation and an iterative method based on Centroidal Voronoi Tessellation (CVT) which acts to minimize the variance in distance between centers of neighboring spots. The process was implemented in MATLAB and within the ASTROID proton treatment planning system to allow for both geometric and dosimetric comparison of the results.
Results: This approach was validated using two different target shapes having concave and convex geometry in homogeneous medium. We quantified the impact of this new approach through direct comparison with grid, contour, and hybrid schemes. Results revealed highly uniform geometric spot distributions with adequate target coverage, improved lateral dose falloff, increased conformity and reduced mean dose to the OAR.
Conclusion: Reduction of spot size and use of movable collimators have been proposed to improve plan quality in proton PBS, however these are not practical to implement on existing machines due to high cost, design complexity and increased treatment times. This work provides an alternative method to improve treatment plan quality that can be used on nearly all existing proton machines with no hardware modifications by using a new scheme that distributes spots conformal to the tumor boundary and uniformly within the target to improve the lateral penumbra in PBS proton therapy treatments. This work will be extended to include spot placement with heterogeneity.
Funding Support, Disclosures, and Conflict of Interest: This work is partially funded by .decimal LLC, Sanford, FL. No conflict of interest