Optimal Delivery of Fluence Profile Using Dynamic Multi-Leaf Collimator Leaf Trajectory Optimization

Abstract

In this paper we show how to employ dynamically generated fluence maps to generate the perfect balance between distribution time and fluence map quality. Our technique is based on a mathematical model that optimizes leaf trajectories and dosage rates for a specific delivery time in order to accurately mimic the desired fluence pattern. We will apply the typical two-step design method to handle the problem of fluence map distribution. The optimization problem is tackled by first optimizing for fluence mapping and then using the leaf-sequencing method to generate precise fluence maps. We formulate dosage rate and leaf trajectory optimization as a non-convex constant optimization problem and use an interior point technique to find the best solution based on arbitrarily adjusted acceptable beginning results. With a prostate example, we put our method to the test and display the resulting balance curves.