Non-linear distortion correcting function for MR diffusion imaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for correcting eddy current-dependent distortions in diffusion-weighted magnetic resonance images are limited by their assumption of affine transformations, which is not always accurate in modern MR systems, leading to incomplete correction and increased computational effort as more complex geometries are considered.
Innovation Solution
A method that uses system-specific distortion correcting functions to account for non-linear transformations, determined through iterative optimization using similarity measures like Normalized Mutual Information, allowing for direct registration of diffusion-weighted images and reducing the need for additional adjustment measurements, while focusing on relevant polynomial elements to maintain computational efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If affine transformations are used for distortion correction, then computational effort is reduced, but correction precision deteriorates because complex interference field geometries are not accounted for
Solution Approach 1:
The distortion correction is segmented into two parts: (1) a linear/affine transformation component that handles the dominant distortion with low computational cost, and (2) a non-linear component that accounts for complex interference field geometries. This segmentation allows the patent to achieve high correction precision without excessive computational complexity by addressing different distortion aspects separately.
Solution Approach 2:
The patent changes the transformation model from purely affine to a combined affine-nonlinear parameterization. By introducing additional non-linear parameters while maintaining the affine foundation, the system achieves more precise distortion correction. The parameter changes are justified by the need to model complex interference field geometries that cannot be captured by affine transformations alone.
2Measurement precision
If more complex geometries are considered in distortion correction, then correction accuracy improves, but computational effort increases significantly
Solution Approach 1:
The patent applies partial non-linear correction rather than full non-linear correction. By combining affine transformations with selective non-linear adjustments, the system achieves sufficient correction accuracy for complex geometries without the full computational burden of complete non-linear modeling. This partial action approach balances accuracy requirements with computational resource constraints.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enables more precise correction of diffusion-weighted images by considering complex interference field geometries beyond affine transformations, improving accuracy and reducing computational costs by limiting the number of correction parameters.
Implementation Method 1
Eddy current fields can disadvantageously be caused by the diffusion gradients, these eddy current fields in turn leading to image distortions
Data Source
AI summary
In a method and magnetic resonance apparatus to reduce distortions in magnetic resonance diffusion imaging, a magnetic resonance data acquisition system is operated to acquire magnetic resonance data in a first measurement with a first diffusion weighting, and to acquire magnetic resonance data in a second measurement with a second, different diffusion weighting. A non-linear, system-specific distortion-correcting function is determined on the basis of system-specific information that is specific to said magnetic resonance data acquisition system. Correction parameters are calculated to correct distortions in subsequently-acquired diffusion-weighted magnetic resonance images, based on the data acquired in the first and second measurements with the system-specific distortion-correcting function applied thereto. The subsequently-acquired diffusion-weighted magnetic resonance images are corrected using the correction parameters to at least reduce distortions therein.