Regularized Decomposition for MRI Fat Suppression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional fat suppression methods in MRI, such as STIR and spectral-spatial pulses, fail to provide reliable and uniform fat suppression in areas with magnetic field heterogeneities, leading to reduced signal-to-noise ratio and limited applications, especially in extremity and large field of view imaging.
Innovation Solution
A method using a regularized decomposition algorithm to combine magnetic resonance image signals, which applies a magnetic resonance excitation, acquires multiple image signals, and iteratively combines them using regularization techniques to enhance noise performance and separate fat from water effectively, even in heterogeneous magnetic fields.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional fat saturation methods are used, then fat suppression is adequate in homogeneous Bo field areas, but fat saturation fails in areas with magnetic field heterogeneities
Solution Approach 1:
The patent changes the parameter of field map smoothing by applying regularization with different smoothing parameters (lambda values) to adapt to varying degrees of field heterogeneity. This allows the method to maintain fat suppression reliability across both homogeneous and heterogeneous magnetic field conditions by adjusting the smoothing parameter according to the specific imaging region
Solution Approach 2:
The patent segments the image into different regions based on field homogeneity characteristics and applies different regularization smoothing parameters to different segments. This allows optimized fat suppression for each region - stronger smoothing for heterogeneous areas and weaker smoothing for homogeneous areas, thereby resolving the contradiction between reliability and adaptability
2Reliability
If STIR imaging is used for uniform fat suppression, then fat suppression is uniform, but signal-to-noise ratio is reduced and contrast is mixed
Solution Approach 1:
Instead of applying uniform fat suppression across the entire image like STIR, the patent applies local quality processing by using spatially varying regularization smoothing parameters. The smoothing parameter lambda is adjusted locally based on the field homogeneity of each region, allowing uniform fat suppression only where needed while preserving SNR in other regions
Solution Approach 2:
The patent substitutes the mechanical T1-based inversion recovery mechanism of STIR with a field map-based iterative decomposition approach. This replacement allows fat suppression to be achieved through mathematical optimization rather than physical T1 weighting, thereby maintaining both uniformity and SNR
3Reliability
If spectral-spatial or water selective pulses are used, then fat suppression is achieved, but the method is sensitive to field inhomogeneities
Solution Approach 1:
The patent introduces a field map as an intermediary that characterizes the magnetic field heterogeneity. By explicitly modeling the field inhomogeneity through the field map and incorporating it into the iterative decomposition process, the method compensates for field variations rather than being sensitive to them, thereby resolving the contradiction between fat suppression effectiveness and field inhomogeneity sensitivity
4Measurement precision
If regularization with high smoothing parameter is used, then noise performance is improved and field maps are smoothed, but bias increases
Solution Approach 1:
The patent makes the regularization smoothing parameter dynamic rather than static. The parameter lambda is adjusted based on local field homogeneity characteristics - higher smoothing is applied in heterogeneous regions where noise reduction is more beneficial, while lower smoothing is applied in homogeneous regions where accuracy is more critical. This dynamic adaptation resolves the contradiction between noise performance and separation accuracy
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 improves noise performance and achieves clinically acceptable bias, allowing for effective fat suppression across various imaging applications, including those with magnetic field inhomogeneities, by utilizing regularization to smooth field maps and optimize signal averaging.
Implementation Method 1
A magnetic resonance imaging excitation is applied
Implementation Method 2
exploit the difference in chemical shifts between water and fat and in order to separate water and fat into separate images
Data Source
AI summary
A method for generating a magnetic resonance images is provided. A magnetic resonance imaging excitation is applied. A plurality of magnetic resonance image signals is acquired. The plurality of image signals is combined iteratively by using a regularized decomposition algorithm. An image created from combining the plurality of image signals iteratively is displayed.