MRI Image Reconstruction Using K-Space Consistency Feedback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing magnetic resonance imaging (MRI) reconstruction methods, particularly those using deep learning, often fail to ensure data consistency, leading to suboptimal image reconstructions, especially in undersampled k-space scenarios.
Innovation Solution
A computer-implemented method utilizing a trained machine learning model (MLM) with a refinement module and optimization modules to optimize MR measurement data through predefined target functions, ensuring data consistency and improving image reconstruction quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If deep learning methods are used for MR image reconstruction, then image quality can be improved, but data consistency is not ensured
Solution Approach 1:
The patent implements a feedback mechanism where the reconstruction is iteratively refined by comparing the current reconstruction with the measured k-space data. The difference between the forward-encoded reconstruction and the actual measurements is used to update the reconstruction, ensuring data consistency while maintaining image quality. This closed-loop feedback ensures that the reconstruction remains consistent with the acquired data throughout the iterative process.
Solution Approach 2:
The patent applies preliminary actions by using a pre-trained machine learning model to generate an initial reconstruction before the iterative refinement process. This initial reconstruction, created by the ML model, serves as a starting point that already incorporates learned patterns and relationships, enabling the subsequent iterative process to focus on refining data consistency rather than building the reconstruction from scratch.
2Reliability
If conventional reconstruction techniques are used, then data consistency is maintained, but image quality is suboptimal
Solution Approach 1:
The patent merges conventional reconstruction techniques with machine learning methods by combining the iterative refinement process (which ensures data consistency) with the machine learning-based initial reconstruction (which provides high image quality). The system integrates the strengths of both approaches: the reliability of conventional methods in maintaining data consistency and the image quality enhancement capabilities of deep learning.
Solution Approach 2:
The patent introduces an intermediary mechanism in the form of a machine learning model that acts as a mediator between the raw k-space data and the final image reconstruction. This intermediary component processes the measurement data and generates an initial reconstruction that balances data consistency with image quality, serving as a bridge between conventional and advanced reconstruction methods.
3Manufacturing precision
If iterative optimization processes are applied, then image quality improves, but computational effort increases
Solution Approach 1:
The patent applies preliminary action by using a pre-trained machine learning model to generate an initial reconstruction before the iterative optimization process. This preliminary reconstruction, created by the ML model, serves as a starting point that already incorporates learned patterns and relationships, enabling the subsequent iterative process to focus on refining data consistency rather than building the reconstruction from scratch, thus reducing the overall computational effort required.
Solution Approach 2:
The patent applies partial action by implementing a limited number of iterative refinement steps rather than exhaustive optimization. The iterative process is designed to perform just enough refinements to achieve data consistency and image quality improvement, stopping before complete convergence would be required. This partial optimization approach significantly reduces computational effort while maintaining acceptable image quality.
Data Source
AI summary
For image reconstruction in magnetic resonance (MR) imaging MR measurement data, which represents an imaged object, is obtained and refined MR data is created by a refinement module of a trained MLM (machine learning model) being applied to module input data dependent on the MR measurement data. An image reconstruction is created depending on the refined MR data, wherein: i) optimized MR data is created depending on the MR measurement data in that, by variation of variable image data, a predefined target function is optimized, and the module input data depends on the optimized MR data; and/or ii) further optimized MR data is created depending on the refined MR data in that, by variation of variable image data, a further target function is optimized, and the image reconstruction is created depending on the further optimized MR data.