Integrated Model for Radiation Therapy Planning

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current knowledge-based medical treatment planning for radiation therapy relies on manual selection of predictive models by experts, which is time-consuming and unreliable, especially when dealing with patient geometries outside the limited regions covered by individual models.

Innovation Solution

An integrated model combining multiple predictive models using a hierarchical clustering algorithm for automatic selection and prediction, enabling coverage of extended effective regions without manual intervention.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If multiple predictive models are used to cover extended regions, then prediction reliability is improved, but manual model selection becomes time-consuming and complex

Engineering Contradiction:
Improveprediction reliabilityVSAvoidmanual model selection time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The system performs automatic model selection through a decision hierarchy that autonomously evaluates patient data features and selects appropriate predictive models without requiring manual expert intervention. The algorithm self-determines which model to apply based on hierarchical rules examining tumor geometry, patient anatomy, and treatment parameters.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The model selection process is segmented into a hierarchical structure with multiple levels, where each level evaluates specific features and narrows down model choices. This segmented approach breaks down the complex selection task into manageable stages, improving both automation and reliability.

Inventive Principle:
Principle #1Segmentation

2Device complexity

If individual predictive models are used with limited regions, then model complexity is reduced, but coverage of patient data features is insufficient

Engineering Contradiction:
Improvemodel complexityVSAvoidcoverage of patient data features
Core Design Contradiction:
Device complexityVSAdaptability or versatility

Solution Approach 1:

The integrated model system achieves universality by combining multiple specialized predictive models into a unified framework. Each individual model maintains its simplicity while the collective system covers extended regions and diverse patient data features through automatic model selection based on hierarchical feature evaluation.

Inventive Principle:
Principle #6Universality (Multi-functionality)

3Measurement precision

If manual expert judgment is used for model selection, then prediction accuracy is maintained, but productivity and efficiency are reduced

Engineering Contradiction:
Improveprediction accuracyVSAvoidtherapy planning efficiency
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The system replaces the mechanical process of manual expert model selection with an automated algorithmic decision hierarchy. The algorithm evaluates patient data features and selects appropriate models automatically, maintaining prediction accuracy while dramatically improving therapy planning efficiency and productivity.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentEP3403694B1Radiation therapy planning using integrated model
Publication Date: 2022.12.28 SIEMENS HEALTHINEERS INTERNATIONAL AG
  • EP3403694B1 patent drawingFigure 1
  • EP3403694B1 patent drawingFigure 2
  • EP3403694B1 patent drawingFigure 3A

AI summary

A method of automatically generating an integrated model for planning a therapy comprises: configuring an input interface 503 operable to receive patient data 501 of a plurality of features; accessing a plurality of predictive models 520, wherein each predictive model is operable to generate a therapy prediction based on said patient data; associating an applicable category with respect to each of said plurality of features for each predictive model; integrating 530 said plurality of predictive models into a hierarchy comprising a bottom level and one or more decision levels, wherein said bottom level comprises said plurality of predictive models, and wherein said one or more decision levels are operable to automatically select one or more predictive models from said bottom level based on applicable categories associated therewith; and configuring an output interface 504 operable to output a therapy prediction generated by a predictive models in response to said patient data.