Neural Network Ghost Artifact Removal in Echo Planar Imaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for removing ghost artifacts in Echo Planar Imaging (EPI) are inefficient, particularly in high-field MRI, due to sensitivity to magnetic field nonuniformity and computational complexity, and often require additional scans or lengthy processing times.
Innovation Solution
A neural network-based approach that interpolates Fourier space data using a low-rank Hankel matrix constraint and convolution framelet, allowing for stable removal of ghost artifacts without a reference scan, by dividing and interpolating odd and even k-space data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional ghost artifact removal methods (reference scan, phase correction) are used, then ghost artifacts can be removed, but processing time increases and additional scans are required
Solution Approach 1:
The patent extracts and removes the harmful phase difference component between even and odd scan lines through complex conjugate multiplication and subtraction operations, isolating the ghost artifact cause and eliminating it without requiring additional reference scans
Solution Approach 2:
The patent uses the relationship between even and odd scan lines to create a virtual reference from the target image itself, copying the necessary information from existing data rather than requiring separate reference acquisitions
2Reliability
If matrix-based restoration methods (low-rank Hankel matrix) are used, then ghost artifacts can be removed, but computational complexity increases significantly
Solution Approach 1:
The patent replaces complex matrix factorization and iterative optimization algorithms with direct algebraic operations including complex conjugate multiplication, element-wise operations, and simple matrix subtractions, dramatically reducing computational burden while maintaining restoration effectiveness
Solution Approach 2:
The patent changes the mathematical approach from solving for matrix rank parameters through iterative optimization to direct calculation using closed-form algebraic expressions, transforming an computationally intensive problem into an efficient analytical solution
3Reliability
If methods sensitive to magnetic field nonuniformity are used, then ghost artifacts can be removed under ideal conditions, but performance degrades in high-field MRI with large nonuniformity
Solution Approach 1:
The patent applies local phase correction by processing each pixel independently through complex conjugate multiplication, allowing the method to adapt to local field variations without being constrained by global assumptions about magnetic field uniformity
Solution Approach 2:
The patent makes the correction method self-adapting to local field conditions by using the actual image data itself to compute correction factors, eliminating the need for separate field mapping scans or assumptions about field uniformity
Data Source
AI summary
Disclosed herein are a method and an apparatus for removing ghost artifacts of an echo planner image using a neural network. An image processing method according to an embodiment of the inventive concept includes receiving Fourier space data of an echo planar image, and restoring the echo planar image in which ghost artifacts are removed using a neural network. The receiving of the Fourier space data may include dividing the Fourier space data into the odd-numbered Fourier space data and even-numbered Fourier space data, and the restoring of the echo planar image may include obtaining the odd-numbered Fourier space data and even-numbered Fourier space data with the Fourier space interpolated using the neural network and restoring the echo planar image in which the ghost artifacts are removed based on the odd-numbered Fourier space data and even-numbered Fourier space data with the Fourier space interpolated.