CNN Correction Network for Abdominal Organ Segmentation in MR-IGART

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

The contouring process for MRI-guided adaptive radiotherapy (MR-IGART) is time-consuming and prone to inter-observer variation, particularly when segmenting abdominal cavity organs like the liver, kidneys, stomach, bowel, and duodenum.

Innovation Solution

A method utilizing a convolutional neural network (CNN) with a correction network to accurately segment organs in 3D MR images, incorporating anatomical constraints to improve segmentation accuracy and efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If traditional manual contouring methods are used for MR-IGART, then segmentation accuracy can be maintained through expert observation, but the process is time-consuming and prone to inter-observer variation

Engineering Contradiction:

Improvecontouring speedVSAvoidsegmentation accuracy

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The system enables automatic self-segmentation of organs through CNN-based networks that process MRI images independently without requiring manual expert contouring, thereby expediting the contouring process while maintaining consistent segmentation results

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent replaces the mechanical manual contouring process with an automated CNN-based image processing system that uses deep learning algorithms to automatically segment organs, eliminating inter-observer variation and significantly reducing contouring time

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If simple CNN segmentation is used, then processing speed is fast, but segmentation accuracy is insufficient for complex abdominal organs

Engineering Contradiction:

Improvesegmentation accuracyVSAvoidnetwork complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The complex segmentation task is divided into multiple stages using separate CNN networks: a first CNN for initial organ segmentation and a second CNN for correcting erroneous labels, allowing each network to focus on specific aspects of the segmentation problem and improve overall accuracy

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The correction network receives feedback from the initial segmentation results and iteratively refines the organ labels by identifying and correcting erroneous segmentations, thereby improving accuracy without requiring excessively complex network architecture

Inventive Principle:

Principle #23Feedback

3Reliability

If anatomical constraints are enforced through explicit rules, then segmentation reliability improves, but system complexity and parameter tuning requirements increase

Engineering Contradiction:

Improvesegmentation reliabilityVSAvoidsystem complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The patent replaces explicit rule-based anatomical constraint enforcement with a data-driven CNN-based correction network that automatically learns and enforces anatomical relationships from training data, improving reliability without requiring manual parameter tuning or complex rule systems

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS12292494B2Methods and systems for segmenting organs in images using a CNN-based correction network

Publication Date: 2025.05.06 WASHINGTON UNIV IN SAINT LOUIS

AI summary

Among the various aspects of the present disclosure is the provision of methods and systems for segmenting images and expediting a contouring process for MRI-guided adaptive radiotherapy (MR-IGART) comprising applying a convolutional neural network (CNN), wherein the CNN accurately segments organs (e.g., the liver, kidneys, stomach, bowel, or duodenum) in 3D MR images.