MRI Shimming via Iterative Weighted Fitting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing MRI shimming techniques are prone to errors from phase unwrapping and are unsuitable for non-standard magnet designs, leading to suboptimal magnetic field uniformity and dependence on regions that cannot be fitted with available shims.
Innovation Solution
A method involving spatial partial derivatives of residue phase maps and shim functions, radial weight generation, regularization factor selection, and iterative fitting to determine shim coefficients, which can accommodate any magnet design and mitigate unwrapping errors by focusing on regions of high certainty first.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If phase unwrapping is performed to determine shim currents, then magnetic field uniformity can be improved, but errors in phase unwrapping propagate and cause large areas of significant error
Solution Approach 1:
The patent segments the magnetic field space into multiple regions with different weighting factors. By dividing the fitting process into region-specific calculations rather than a global calculation, errors in phase unwrapping are contained within local regions and do not propagate across the entire field of view. Each region can be independently optimized with appropriate weighting.
Solution Approach 2:
The patent applies local quality by using spatially varying weighting factors that assign different importance to different regions of the magnetic field. Regions with high signal-to-noise ratio and accurate phase unwrapping receive higher weights, while regions prone to errors receive lower weights. This local differentiation prevents error propagation from affecting the overall field uniformity.
2Ease of manufacture
If specific polynomial sets are used for determining optimal shim currents, then calculation simplicity is maintained, but the method becomes unsuitable for non-standard magnet designs
Solution Approach 1:
The patent creates a universal shimming method that can accommodate multiple magnet designs including both standard and non-standard configurations. By using a weighted least-squares fitting approach with region-specific weighting factors rather than relying on specific polynomial sets, the method becomes adaptable to various magnet geometries and shim coil arrangements while maintaining computational efficiency.
Solution Approach 2:
The patent changes the approach from fixed polynomial sets to a flexible weighted fitting method where weighting factors can be adjusted based on the specific magnet design, signal-to-noise characteristics, and region importance. This parameter flexibility allows the same fundamental method to work across different magnet configurations without requiring design-specific polynomial developments.
3Productivity
If fitting is performed across all regions equally, then computational efficiency is maintained, but regions with high uncertainty dominate the error and reduce overall precision
Solution Approach 1:
The patent applies local quality by assigning different weighting factors to different spatial regions based on their signal-to-noise ratio and phase unwrapping reliability. Regions with high uncertainty receive lower weights, preventing them from dominating the fitting process, while regions with high confidence receive higher weights to improve overall fitting accuracy. This localized weighting maintains computational efficiency while significantly improving precision.
4Manufacturing precision
If iterative refinement is applied to improve shim coefficient accuracy, then magnetic field uniformity increases, but computation time increases
Solution Approach 1:
The patent performs preliminary actions by pre-calculating optimal weighting factors based on initial signal-to-noise assessments and phase unwrapping quality metrics before the main fitting process. This preliminary weighting setup enables the iterative refinement to converge faster by preventing error propagation from the outset, thereby reducing the number of iterations needed to achieve the desired field uniformity and cutting overall computation time.
Data Source
AI summary
To achieve a uniform magnetic field in an MRI system, fitting can be performed using a partially differentiated residue phase map, differentiated shim functions, radial weights, a regularization factor, a discontinuity mask, and/or a signal intensity mask to determine coefficients for shim functions. The fitting can be performed iteratively, where the regularization factor is stronger and the radial weights focus on areas of higher confidence during earlier iterations. During later iterations, the regularization factor gradually gets weaker and the radial weights gradually focus on areas of lower confidence.