Knowledge-Based Medical Image Segmentation via Deformable Registration

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current medical image automatic segmentation methods are inefficient due to the need for manual seeding and editing, especially with increasing medical imaging resolution, and are often structure-specific, leading to prolonged treatment planning times and errors.

Innovation Solution

A knowledge-based system using a plurality of reference image sets with segmented and contoured structures, employing deformable image registration and weighted averaging to generate accurate segmentations, reducing manual intervention and improving precision across various organs.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual seeding and editing is used for automatic segmentation, then structure-specific accuracy can be maintained, but treatment planning time increases significantly

Engineering Contradiction:
Improvesegmentation accuracyVSAvoidtreatment planning time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The system performs preliminary segmentation using deformable image registration and statistical models before manual editing, preparing an initial accurate segmentation that reduces the subsequent manual work required. Reference image sets are pre-processed and stored for rapid retrieval during treatment planning.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system creates copies of reference image sets with segmented structures and applies deformable registration to adapt them to the current patient's anatomy. This allows automatic generation of segmentation based on copied and transformed reference data, reducing manual seeding requirements.

Inventive Principle:
Principle #26Copying

2Measurement precision

If structure-specific automatic contouring software is used, then segmentation accuracy for specific organs improves, but the system complexity increases and requires multiple specialized tools

Engineering Contradiction:
Improveorgan segmentation accuracyVSAvoidsoftware system complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system implements a universal deformable image registration framework that can handle multiple organ types and imaging modalities simultaneously. The statistical model and reference image sets are adapted to different structures through the same core algorithm, eliminating the need for multiple specialized software tools.

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

Solution Approach 2:

The system adjusts parameters of the deformable registration model and statistical features to optimize segmentation for different organ types. By changing model parameters rather than using different software, the system maintains simplicity while achieving structure-specific accuracy.

Inventive Principle:
Principle #35Parameter changes

3Ease of operation

If conventional graphics programs are used for manual contour tracing, then flexibility in editing is maintained, but productivity decreases and human error increases

Engineering Contradiction:
Improvecontour editing flexibilityVSAvoidsegmentation speed
Core Design Contradiction:
Ease of operationVSProductivity

Solution Approach 1:

The system performs automatic segmentation and contour generation without requiring manual tracing, making the system self-sufficient for the primary segmentation task. This eliminates the time-consuming manual contour tracing while maintaining the ability to edit results when needed.

Inventive Principle:
Principle #25Self-service

4Measurement precision

If increasing medical imaging resolution is implemented, then diagnostic accuracy improves, but the amount of data to be segmented increases, prolonging processing time

Engineering Contradiction:
Improvediagnostic accuracyVSAvoidsegmentation processing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The system performs preliminary deformable registration and statistical analysis on high-resolution images to generate an initial segmentation framework before detailed contouring, reducing the time required to process the large amount of data from high-resolution imaging.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system extracts key statistical features and landmark points from high-resolution images for the deformable registration process, working with a reduced set of critical data points rather than processing every pixel, thereby maintaining diagnostic accuracy while reducing processing time.

Inventive Principle:
Principle #2Taking out (Extraction)

Data Source

PatentEP3742393B1Knowledge-based automatic image segmentation
Publication Date: 2024.09.18 VARIAN MEDICAL SYSTEMS INC
  • EP3742393B1 patent drawingFigure 1
  • EP3742393B1 patent drawingFigure 2
  • EP3742393B1 patent drawingFigure 3

AI summary

A method for medical image segmentation. The method includes accessing and updating a knowledge-base in accordance with embodiments of the present invention. The techniques include: receiving a medical image and computing a sparse landmark signature based on the medical image content. Next, a knowledge-base is searched for representative matches to form a base set, wherein the base set comprises a plurality of reference image sets. A portion of the plurality of reference image sets of the base set is deformed to generate mappings from the base set to the medical image set. Finally a weighted average segmentation for each structure of interest of the medical image set is determined.