Radiotherapy Fluence Map Optimization Using Anatomy Projections
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The process of creating radiation therapy treatment plans is time-consuming and subjective, often requiring trial-and-error adjustments to balance target dose versus organ-at-risk sparing, especially with complex anatomies involving multiple critical organs, leading to uncertainty in plan quality.
Innovation Solution
Utilizing a trained machine learning model, such as an artificial neural network, to generate fluence maps based on patient anatomy projections, optimizing fluence distributions for improved radiotherapy planning and reducing the need for manual adjustments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If traditional offline treatment planning with manual trial-and-error optimization is used, then the treatment plan can be customized for each patient, but the process is time-consuming and subjective
Solution Approach 1:
The system performs preliminary actions by pre-training the neural network model on extensive treatment plan data before actual treatment planning. This pre-computed knowledge enables rapid generation of accurate fluence maps during clinical use, eliminating the need for time-consuming manual trial-and-error optimization while maintaining high plan quality
Solution Approach 2:
The patent replaces the mechanical trial-and-error optimization process with an artificial neural network-based system. The neural network learns complex optimization patterns from training data and automatically generates optimized fluence maps, substituting manual planner efforts with automated intelligent computation that is both faster and more consistent
2Adaptability or versatility
If the number of organs-at-risk increases, then the treatment plan complexity increases, but the difficulty of sparing critical organs increases significantly
Solution Approach 1:
The neural network transforms the complex multi-constraint optimization problem into a parameter estimation problem. By learning from training data containing multiple organs-at-risk scenarios, the network internalizes complex spatial relationships and dose constraints, enabling it to handle complex anatomies through parameter optimization rather than explicit constraint satisfaction
Solution Approach 2:
The patent introduces anatomy projection images as an intermediary representation between the complex 3D anatomical data and the fluence map generation process. This 2D projection view simplifies the input complexity while preserving essential anatomical information, making the problem more tractable for the neural network to process
3Reliability
If manual treatment planning is used, then the planner can apply clinical judgment, but the process is subjective and leads to uncertainty in plan quality
Solution Approach 1:
The system incorporates feedback mechanisms through the neural network's learning process, where the model continuously improves by learning from training data containing optimal treatment plans. This feedback loop enables the system to capture and reproduce expert clinical judgment patterns, providing consistent and reliable plan quality without subjective variation
Solution Approach 2:
The neural network performs self-service by automatically generating optimized fluence maps without requiring manual intervention. The system serves itself by using the trained model to directly produce treatment plans, eliminating the need for planners to repeatedly apply subjective judgment while maintaining consistency across all cases
Data Source
Figure 1
Figure 2A
Figure 2B
AI summary
Systems and methods are disclosed for generating fluence maps for a radiotherapy treatment plan that uses machine learning prediction. The systems and methods include identifying image data that indicates treatment constraints for target dose areas and organs at risk areas in an anatomy of the subject, generating anatomy projection images that represent a view of the subject from respective beam angles, using a trained neural network model to generate the computer-simulated fluence map representations based on the anatomy projection images, where the fluence maps indicate a fluence distribution of the radiotherapy treatment at each of the beam angles.