MRI Streak-Artifact Suppression via Adaptive Coil Weighting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Abdominal MRI images suffer from streaking artifacts due to gradient nonlinearities, particularly from unsuppressed fat, which can obscure pathology and lead to unacceptable loss in image quality, and existing methods to mitigate these artifacts often result in signal-to-noise ratio loss.
Innovation Solution
A method involving the generation of an interference covariance matrix from coil images to determine a coil weight vector, which is used to produce a streak-suppressed MR image as a weighted sum of the coil images, effectively reducing streaking artifacts while minimizing signal loss.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If coil removal methods are used to prune coils contributing to streaking artifacts, then artifact reduction is achieved, but signal-to-noise ratio is lost
Solution Approach 1:
The patent applies local quality by computing spatially-varying coil weight maps W(x,y) for each image pixel location. Each pixel receives optimized weights tailored to its local characteristics, allowing artifact suppression in artifact-prone regions while preserving signal in high-SNR regions. This resolves the contradiction by making the coil combination adaptive to local image quality requirements rather than applying uniform pruning.
Solution Approach 2:
The patent implements dynamics by using data-driven, adaptive coil weighting that varies spatially across the image. The weight maps W(x,y) are dynamically determined from the acquired coil data itself through covariance analysis, rather than being fixed or manually selected. This allows the system to automatically adjust the contribution of each coil based on local artifact levels and signal quality, resolving the trade-off between artifact reduction and SNR preservation.
2Object-affected harmful factors
If manual identification of problematic coils is performed, then artifact reduction is achieved, but labor and time are consumed
Solution Approach 1:
The patent applies self-service by implementing an automated algorithm that identifies and weights coils based on their contribution to streaking artifacts without manual intervention. The system uses covariance analysis of the acquired coil data to automatically determine which coils contribute to artifacts and computes optimal weights accordingly. This eliminates the need for laborious manual identification while achieving effective artifact suppression.
Solution Approach 2:
The patent implements feedback by using the acquired coil data itself to guide the artifact suppression process. The covariance matrix and weight maps are computed from the actual measured signals, creating a closed-loop system where the data informs the processing parameters. This automated feedback mechanism replaces manual inspection and coil selection, significantly reducing processing time while maintaining effectiveness.
3Object-affected harmful factors
If coils are heavily down-weighted to minimize artifacts, then artifact reduction is achieved, but signal-to-noise ratio is lost
Solution Approach 1:
The patent prevents information loss by applying local quality through spatially-varying weights W(x,y). In regions where coils contribute to artifacts, their weights are reduced; in regions where they provide valuable signal, their weights are maintained or enhanced. This localized approach ensures that signal intensity is preserved where needed while suppressing artifacts where they occur, avoiding the blanket signal loss that results from heavy down-weighting.
Solution Approach 2:
The patent applies parameter changes by dynamically adjusting coil weights based on local image characteristics rather than using fixed heavy down-weighting. The covariance-based analysis identifies regions where artifact suppression is needed and modifies weights accordingly, preserving signal intensity in regions where the parameter change (weight adjustment) is not necessary for artifact reduction.
Data Source
AI summary
A method for producing a streak-suppressed magnetic resonance (MR) image of a subject includes generating an interference covariance matrix {circumflex over ( )}R front N coil images Ij (x,y), j={1, 2, . . . , N}, each of the N coil images Ij (x,y) corresponding to MR signals detected by a respective one of a phased array of N coils of an MRI scanner. The MR signals originate in voxels of the subject corresponding to an artifact-corrupted region of a coil image. Coordinates (x,y) correspond to a location within a cross-sectional plane of the subject. The method also includes, for subject-regions of cross-sectional plane centered at a respective location (x,y), determining a coil weight vector W (x,y) from {circumflex over ( )}R. The method also includes generating the streak-suppressed MR image as a weighted sum of the coil images Ij (x,y), each weight of the weighted sum being Wj* (x,y), a jth element of a complex conjugate of coil weight vector W (x,y).