Machine Learning Model for Virtual Bolus Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for creating virtual boluses in radiation therapy are manual, time-consuming, and unreliable, relying on subjective medical professional skills, which fail to accurately account for patient movement and breathing during treatment planning.
Innovation Solution
A machine-learning model is trained using historical patient data, including medical images and treatment parameters, to predict virtual bolus attributes, allowing for automatic and efficient generation of treatment plans that adapt to new data distributions, reducing reliance on human intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual methods are used to create virtual boluses, then medical professionals can exercise control over the treatment plan, but the process becomes time-consuming and unreliable
Solution Approach 1:
The machine-learning model performs self-service by automatically generating virtual bolus attributes without requiring manual intervention from medical professionals. The system uses historical data and trained algorithms to autonomously create treatment plans, eliminating the time-consuming manual process while maintaining reliability through consistent algorithmic application.
Solution Approach 2:
The patent replaces the mechanical manual process of virtual bolus creation with an automated machine-learning system. The mechanical system of manual drawing and adjustment is substituted by computational algorithms that process medical images and generate virtual bolus attributes automatically, significantly reducing time while improving reliability.
2Reliability
If manual methods are used to create virtual boluses, then medical professionals can adjust parameters subjectively, but the results become inconsistent and dependent on individual skills
Solution Approach 1:
The system changes the parameters of virtual bolus creation from subjective manual adjustments to objective algorithmic outputs. The machine-learning model transforms qualitative professional judgment into quantitative, reproducible parameters based on trained patterns from historical data, ensuring consistency while managing complexity through standardized computational processes.
Solution Approach 2:
The machine-learning model incorporates feedback mechanisms by continuously learning from historical treatment data and outcomes. This feedback loop allows the system to refine its predictions and improve consistency over time, replacing individual professional variability with a self-improving algorithmic system that maintains stable, reproducible results.
3Ease of operation
If conventional methods are used to account for patient movement, then physical boluses or manual MLC modification are required, but the process becomes tedious and subjective
Solution Approach 1:
The machine-learning model provides self-service by automatically accounting for patient movement and breathing patterns without requiring manual bolus placement or MLC modification. The system processes medical images and treatment history to autonomously generate adjusted treatment plans, making the process easy to operate while significantly improving productivity through automation.
Solution Approach 2:
The patent replaces mechanical physical boluses and manual MLC adjustments with a computational machine-learning system. This substitution eliminates the tedious manual processes of physical bolus placement and subjective parameter adjustment, enabling automated accounting for patient movement while dramatically improving treatment planning efficiency.
Data Source
AI summary
Disclosed herein are methods and systems for predicting a virtual bolus in order to generate a radiation therapy treatment plan comprising training, by a processor, a machine-learning model using a training dataset comprising a set of medical images corresponding to a set of previously performed radiation therapy treatments, each medical image comprising at least one planning target volume and a non-anatomical region added to the medical image; and executing, by the processor, the machine-learning model using a medical image not included within the training dataset, the machine-learning model predicting an attribute of a non-anatomical region for the medical image not included in the training dataset.