MRI Compressed Sensing Location-Dependent Regularization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The challenge in compressed sensing magnetic resonance (MR) imaging is finding an optimal regularization parameter that balances noise reduction and anatomical detail preservation, especially in orthopedic images, where the parameter varies significantly with measurement setups and body parts, leading to trial-and-error approaches.
Innovation Solution
A method using location-dependent sensitivity maps to determine a case-specific regularization parameter, which is proportional to noise and inversely proportional to sensitivity, allowing for adaptive image generation that improves signal-to-noise ratio without losing anatomical details, and averaging parameters for regions to maintain consistent image quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If a high regularization parameter is used to reduce noise, then noise in the image is strongly reduced, but important relevant details of the anatomy are lost
Solution Approach 1:
The patent applies local quality by determining location-dependent regularization parameters for different regions of the image. Instead of using a single global regularization parameter, the system calculates different parameter values based on local signal-to-noise ratios in different anatomical regions. This allows noise reduction in low-signal areas while preserving anatomical details in high-signal areas, resolving the contradiction between noise reduction and detail preservation.
2Loss of information
If a low regularization parameter is used to preserve anatomical details, then anatomical structures are maintained, but noise covers the anatomy making identification difficult
Solution Approach 1:
By implementing location-dependent regularization parameters, the system adapts the noise reduction strength to local conditions. In regions with high signal-to-noise ratio, lower regularization parameters preserve anatomical details, while in low signal-to-noise ratio regions, higher parameters reduce noise. This local adaptation resolves the contradiction by allowing both noise reduction and detail preservation in different image regions simultaneously.
3Ease of operation
If a fixed regularization parameter is preselected for different measurements, then the measurement process is simplified, but image quality varies significantly across different body parts and measurement setups
Solution Approach 1:
The system implements self-service by automatically determining the optimal regularization parameters based on the actual measurement data and local signal-to-noise ratios. Instead of requiring manual preselection of parameters for different body parts and measurement setups, the system calculates location-dependent parameters automatically during the reconstruction process. This resolves the contradiction by maintaining ease of operation while achieving consistent high image quality across different measurements through adaptive parameter selection.
4Manufacturing precision
If trial and error is used to find the best regularization parameter, then image quality can be optimized, but time and effort are wasted
Solution Approach 1:
The system applies preliminary action by calculating location-dependent regularization parameters automatically during the reconstruction process based on local signal-to-noise ratios. Instead of requiring iterative trial and error after image reconstruction, the optimal parameters are determined as part of the reconstruction process itself. This resolves the contradiction by achieving optimized image quality without the time loss associated with manual trial and error parameter tuning.
Data Source
Figure 1
Figure 2~3
AI summary
The invention relates to a method for generating a signal-to-noise improved MR image of an object under examination in an MR system using a compressed sensing technology, the method comprising: - determining a first MR signal data set of the object under examination in which a corresponding k-space is randomly subsampled, - determining a location dependent sensitivity map for each of at least one receiving coil used to detect MR signals of the first MR signal data set in the location where the object under examination is located, - determining the signal-to-noise improved MR image using an optimization process of the compressed sensing technology in which a location dependent regularization parameter λ is used, wherein the location dependent regularization parameter λ is determined based on the location dependent sensitivity map.