Two-Stage MRI Motion Correction for Undersampled Image Reconstruction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing retrospective motion compensation techniques in magnetic resonance imaging (MRI) face challenges in maintaining image quality, robustness, and reproducibility, especially when highly undersampled data acquisition is used, leading to issues like noise amplification and aliasing artifacts.

Innovation Solution

A computer-implemented method involving multiple motion correction steps with intermediate image quality improvement using algorithms, particularly trained machine learning models, to enhance the quality, robustness, and reproducibility of MRI images by compensating for motion and sparse k-space sampling.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If retrospective motion compensation is applied to highly undersampled MRI data, then image quality and reproducibility are improved, but processing complexity and computational cost increase

Engineering Contradiction:
Improveimage qualityVSAvoidprocessing complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The motion compensation process is divided into multiple sequential steps: initial motion correction, intermediate image quality improvement, and final optimization. This segmentation allows each step to focus on specific aspects of the problem, reducing overall processing complexity while maintaining reliability

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Image quality improvement algorithms are applied as an intermediate step before final motion correction optimization. This preliminary action prepares the data by reducing noise and artifacts early in the process, making subsequent optimization more efficient and less computationally intensive

Inventive Principle:
Principle #10Preliminary action

2Productivity

If faster imaging sequences with sparse k-space sampling are used, then scan time is reduced, but image quality and reproducibility degrade

Engineering Contradiction:
Improvescan speedVSAvoidimage quality
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

Image quality improvement algorithms serve as an intermediary between sparse k-space sampling and final image reconstruction. These algorithms process the undersampled data to compensate for missing information, allowing fast scanning while maintaining image quality and reproducibility

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The system dynamically adjusts processing parameters based on the degree of undersampling and motion characteristics. By changing parameters such as regularization strength and optimization iterations, the system maintains image quality across different scanning speeds

Inventive Principle:
Principle #35Parameter changes

3Loss of time

If motion correction optimization is performed without intermediate quality improvement, then processing time is reduced, but image quality and robustness deteriorate

Engineering Contradiction:
Improveprocessing timeVSAvoidimage quality
Core Design Contradiction:
Loss of timeVSReliability

Solution Approach 1:

The method applies image quality improvement as a preliminary action before final motion correction optimization. This intermediate step reduces noise and artifacts early, making the subsequent optimization faster and more efficient, thus not increasing overall processing time while improving image quality

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4624972A1Method for providing a final motion corrected image dataset based on magnetic resonance data, data processing system, computer program and computer-readable medium
Publication Date: 2025.10.01 SIEMENS HEALTHINEERS AG
  • EP4624972A1 patent drawingFigure 1
  • EP4624972A1 patent drawingFigure 2
  • EP4624972A1 patent drawingFigure 3

AI summary

Computer-implemented method for providing a final motion corrected image dataset (1) based on magnetic resonance data (2) concerning an object and/or a person (3), in particular a patient, comprising the steps of - receiving magnetic resonance data (2), - determining a first motion corrected image dataset (4) by solving a first optimization problem (5), wherein the first optimization problem (4) depends on the magnetic resonance data (2) and on motion data (6), wherein the motion data (6) concerns a movement of the object and/or the person (3) during the acquisition of the magnetic resonance data (2), - processing the first motion corrected image dataset (4) by an algorithm for image quality improvement (7) to provide a processed image dataset (8), - determining a second motion corrected image dataset (9) by solving a second optimization problem (10) that depends on the magnetic resonance data (2), on the motion data (6) and on the processed image dataset (8), and - either providing the second motion corrected image dataset (9) as the final motion corrected image dataset (1) or determining the provided final motion corrected image dataset (1) based on the second motion corrected image dataset (9).