Machine Learning Artifact Correction in Multi-Shot Echo Planar Imaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
MRI data acquisition is limited by slow data acquisition rates, leading to increased sensitivity to patient motion-induced image artifacts and reduced patient throughput, especially in techniques like multi-shot echo planar imaging (MS-EPI) where physiological motion causes severe image distortion.
Innovation Solution
Implementing a machine learning algorithm, such as a Residual Convolutional Neural Network (CNN), to estimate and remove physiological artifacts from MRI data, allowing for faster and more efficient data acquisition with high isotropic resolution without the need for additional navigator scans, which can compromise efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If rapid imaging sequences like multi-shot echo planar imaging are used to reduce acquisition time, then productivity is improved, but image quality deteriorates due to severe physiological motion artifacts
Solution Approach 1:
The patent applies machine learning algorithms to convert the harmful physiological motion artifacts into beneficial information. The system learns the characteristic patterns of motion artifacts from training data and uses this knowledge to identify and correct similar artifacts in new images, transforming what was previously pure noise into useful correction signals that improve image quality while maintaining fast acquisition speeds
Solution Approach 2:
The patent introduces machine learning models as an intermediary between the raw corrupted images and the final reconstructed images. These models act as mediators that learn the complex relationship between motion artifacts and underlying anatomical structures, enabling accurate artifact removal without requiring direct physical measurements or additional scanning time
2Manufacturing precision
If traditional artifact removal methods using navigator scans are used, then image quality is improved, but productivity deteriorates due to additional scan time requirements
Solution Approach 1:
The patent extracts the artifact removal function from the physical scanning process itself. Instead of using additional navigator scans during the imaging sequence, the system extracts and removes motion artifacts through post-processing machine learning algorithms, separating the diagnostic image acquisition from the artifact correction functions to maintain high acquisition speed while improving image quality
Solution Approach 2:
The patent replaces the mechanical/physical navigator scan system with a computational machine learning system. Rather than physically acquiring additional reference data during scanning, the system uses trained algorithms to computationally estimate and remove artifacts, substituting a complex physical measurement system with a more efficient computational approach
3Productivity
If thick slices with gaps are used to reduce acquisition time, then productivity is improved, but measurement precision deteriorates due to missed information
Solution Approach 1:
The patent performs preliminary machine learning-based artifact removal and image reconstruction before final image analysis and diagnosis. By pre-processing the images to remove artifacts and improve quality, the system enables accurate measurement and analysis from the acquired data without requiring additional scanning or thicker slices, ensuring complete information is captured and preserved
Data Source
AI summary
Systems and methods are provided for improving MRI data acquisition efficiency while providing more detailed information with high resolution and isotropic resolution without gaps. Improved data acquisition efficiency may be achieved by implementing a machine learning algorithm with a hardware processor and a memory to estimate imperfections in fast imaging sequences, such as a multi-shot echo planar imaging (MS-EPI) sequence. These imperfections, such as patient motion, physiological noise, and phase variations, may be difficult to model or otherwise estimate using standard physics-based reconstructions.