MRE Quality Prediction Using Deep Learning During Data Acquisition

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Magnetic resonance elastography (MRE) images often result in poor diagnostic quality due to physiological and mechanical factors, leading to reduced image confidence and the need for reacquisition, necessitating improvements in imaging and measurement techniques.

Innovation Solution

Utilizing deep learning (DL) for automated MRE quality control through AI classification models like ResNet18, ResNet34, ResNet50, SqueezeNet, and MobileNetV2 to assess image quality and identify artifact sources, and AI segmentation models to determine measurable stiffness areas, thereby improving diagnostic efficacy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If MRE imaging is performed to measure tissue stiffness, then diagnostic information about tissue mechanical properties is obtained, but poor diagnostic quality results due to physiological and mechanical factors lead to reduced image confidence and require reacquisition

Engineering Contradiction:

Improvediagnostic qualityVSAvoidreacquisition time

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The patent applies preliminary action by performing quality assessment during the data acquisition process itself. The deep learning model analyzes incoming MRE data in real-time to predict final image quality before the complete acquisition is finished. This allows operators to detect potential quality issues early and take corrective actions during the procedure, preventing wasted reacquisition time while maintaining high diagnostic quality standards

Inventive Principle:

Principle #10Preliminary action

2Measurement precision

If manual quality assessment is performed by operators, then diagnostic quality evaluation is achieved, but inter-observer variability reduces consistency and increases processing time

Engineering Contradiction:

Improvequality assessment consistencyVSAvoidprocessing time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent replaces the mechanical system of manual human observation and assessment with an automated deep learning-based image processing system. The trained neural network model consistently evaluates MRE image quality metrics without inter-observer variability, providing precise and reproducible measurements. This automated system processes images faster than manual review while eliminating human subjectivity, thereby improving both measurement precision and reducing processing time

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system enables self-service by allowing the MRE imaging system to automatically assess its own output quality through integrated deep learning models. The quality assessment functionality is built into the imaging workflow, enabling the system to autonomously evaluate diagnostic quality without requiring external manual intervention. This self-assessment capability maintains consistency while accelerating the overall process

Inventive Principle:

Principle #25Self-service

Data Source

PatentUS20250363631A1Deep learning-based diagnostic quality prediction during magnetic resonance elastography data acquisition

Publication Date: 2025.11.27 GEORGIA TECH RES CORP

AI summary

Disclosed are systems and method for automated magnetic resonance elastography (MRE) quality control and stiffness measurements. The exemplary systems and methods described herein utilize deep learning (DL) to reduce inter-observer variability, improve processing time and workflow constraints, and assist operators with troubleshooting based on artifact sources.