Fluence Map Optimization Using Accelerated Proximal Gradient
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for fluence map optimization in radiation treatment planning, such as interior point methods and gradient-based methods, face limitations in handling large-scale problems and nondifferentiable objective functions, leading to computational intensity and restricted treatment plan quality.
Innovation Solution
The implementation of a proximal gradient method, specifically an accelerated proximal gradient method like FISTA, with smoothed-out nondifferentiable penalty functions to compute fluence maps that ensure precise radiation delivery to target regions while minimizing exposure to organs-at-risk, using penalty functions like Li-type or L2-type penalties.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If interior point methods are used to solve fluence map optimization problems, then small and medium-size problems can be solved effectively, but large-scale problems become prohibitively computationally intensive
Solution Approach 1:
The patent transforms the fluence map optimization problem by changing the parameter representation and objective function formulation. Instead of using traditional linear or quadratic programming formulations that require solving large linear systems, the patent reformulates the problem with a different objective function that can be solved using gradient-based methods, thereby improving computational efficiency for large-scale problems while maintaining optimization accuracy
Solution Approach 2:
The patent replaces the mechanical/mathematical system of interior point methods (which require solving large linear systems of equations) with a gradient-based optimization approach. This substitution eliminates the need to solve large linear systems at each iteration, significantly reducing computational intensity for large-scale fluence map optimization problems
2Productivity
If gradient-based methods are used for fluence map optimization, then computational efficiency improves for large-scale problems, but the ability to handle nondifferentiable objective functions and complex constraints is lost
Solution Approach 1:
The patent changes the parameter formulation of the objective function to make it differentiable while preserving the essential optimization goals. By reformulating nondifferentiable penalty functions (such as those based on L1 norms or other non-smooth constraints) into differentiable forms, the patent enables the use of gradient-based methods without losing the ability to handle complex constraints and objective functions
Solution Approach 2:
The patent introduces intermediary differentiable functions that approximate nondifferentiable objective functions and constraints. These intermediary functions serve as mediators that allow gradient-based methods to operate effectively while still capturing the essential behavior of the original nondifferentiable problem, thereby maintaining handling capability for complex constraints
Data Source
Figure 1A
Figure 1B
Figure 1C
AI summary
Described herein are methods for fluence map generation or fluence map optimization (FMO) for radiation therapy. One variation of a method for generating a fluence map comprises smoothing out nondifferentiable penalty functions and using an accelerated proximal gradient method (e.g., FISTA) to compute a fluence map that may be used by a radiotherapy system to apply a selected dose of radiation to one or more regions of interest (ROI) or volumes of interest (VOI).