SENSE MRI Artifact Removal via Motion Likelihood Map
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current SENSE magnetic resonance imaging techniques face challenges in efficiently removing motion artifacts due to their numerical intensity and the difficulty in identifying and correcting ghosting artifacts, which affects image quality.
Innovation Solution
A method that utilizes a motion likelihood map and anatomical models to identify potential artifact origins, incorporating prior knowledge to accelerate the numerical process and construct an extended SENSE equation with additional columns in the sensitivity matrix to suppress or eliminate ghosting artifacts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If extended SENSE formulation is used to remove ghosting artifacts, then artifact removal capability is improved, but numerical intensity increases
Solution Approach 1:
The patent applies preliminary action by using a motion likelihood map to pre-identify potential artifact origins before performing the extended SENSE reconstruction. This pre-processing step guides the numerical algorithm to focus only on regions with high motion probability, thereby reducing the overall numerical intensity while maintaining effective artifact removal capability.
Solution Approach 2:
The patent implements local quality by applying the extended SENSE formulation selectively only to regions identified by the motion likelihood map as having high probability of containing artifacts. Instead of processing the entire image uniformly, the method concentrates computational resources on local regions where artifacts are most likely to occur, thus reducing global numerical intensity while preserving local artifact removal effectiveness.
2Measurement precision
If numerical methods are used to identify artifact origins, then artifact detection accuracy is improved, but computational burden increases
Solution Approach 1:
The patent introduces a motion likelihood map as an intermediary that bridges the gap between raw image data and artifact origin identification. This intermediary structure pre-processes the image to highlight regions with high motion probability, thereby guiding the subsequent numerical identification process. The motion likelihood map acts as a filter that reduces the search space for artifact origins, improving detection accuracy while reducing computational burden.
3Reliability
If extended SENSE equation is constructed with additional columns, then ghosting artifact suppression is improved, but device complexity increases
Solution Approach 1:
The patent applies local quality by adding columns to the sensitivity matrix selectively based on the motion likelihood map. Instead of uniformly increasing matrix complexity across the entire image, the method adds matrix columns only for regions identified as having high probability of containing artifacts. This localized approach to matrix extension maintains effective ghosting suppression in critical regions while minimizing overall device complexity.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention provides for a magnetic resonance imaging system (100, 300) comprising: a radio-frequency system (116, 122, 124, 126, 126', 126'', 126''') for acquiring magnetic resonance data (152) from an imaging zone (108), wherein the radio-frequency system comprises multiple antenna elements (126, 126', 126'', 126'''); a memory (140) containing machine executable instructions (170) and pulse sequence commands (150), wherein the pulse sequence commands cause the processor to acquire magnetic resonance data from the multiple antenna elements according to a SENSE protocol; and a processor. Execution of the machine executable instructions causes the processor to: control (200) the magnetic resonance imaging system with the pulse sequence commands to acquire the magnetic resonance data; reconstruct (202) a preliminary image (154) using the magnetic resonance imaging data; calculate (204) a fit (159) between an anatomical model (156) and the preliminary image, wherein the anatomical model comprises a motion likelihood map (158); identify (206) at least one image artifact origin (160) at least partially using the motion likelihood map and the fit; determine (208) an extended SENSE equation (162) at least partially using at least one image artifact origin; and construct (210) a corrected SENSE image (164) using the extended SENSE equation.