Multi-contrast MR Image Reconstruction via Sparse Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for reconstructing magnetic resonance (MR) images from undersampled k-space data, such as parallel imaging and compressed sensing, suffer from low signal-to-noise ratio and residual aliasing artifacts, particularly when high acceleration factors are used, compromising the diagnostic value of multi-contrast imaging.
Innovation Solution
A method that reconstructs multi-contrast MR images by using a sparse image coding procedure involving a data consistency iteration step and a denoising iteration step, which includes performing a 2D/3D block matching operation to identify similar patches across reconstructed images and using these patches in a sparsifying operation to provide sparse representations, thereby enhancing image quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If parallel imaging techniques are used to reconstruct MR images from undersampled k-space data, then the scan time is reduced, but the signal-to-noise ratio becomes low
Solution Approach 1:
The patent combines multiple contrast images (T1-weighted, T2-weighted, etc.) into a unified reconstruction framework, merging their respective k-space data and exploiting inter-contrast redundancies. This merging approach allows the system to maintain higher signal-to-noise ratio by pooling information across multiple contrasts while still achieving accelerated scan times through undersampling.
Solution Approach 2:
The patent employs a composite reconstruction strategy that integrates multiple imaging modalities and contrast types into a single reconstruction pipeline. By treating different contrast images as composite data sources with complementary information, the system achieves both time reduction and noise suppression that neither parallel imaging nor compressed sensing can achieve alone.
2Loss of time
If compressed sensing techniques are used to reconstruct MR images from undersampled k-space data, then the scan time is reduced, but residual aliasing artifacts remain
Solution Approach 1:
The patent merges multiple contrast images in the reconstruction process, allowing aliasing artifacts to be suppressed through inter-contrast correlation. By combining information from T1-weighted, T2-weighted, and other contrast images, the system can distinguish true anatomical structures from aliasing artifacts, effectively removing the harmful artifacts while maintaining accelerated scanning.
Solution Approach 2:
The patent converts the presence of aliasing artifacts into a beneficial signal by exploiting the fact that true anatomical structures appear consistently across multiple contrasts while artifacts do not. This allows the reconstruction algorithm to identify and eliminate artifacts, transforming the harmful aliasing effect into useful information for artifact suppression.
3Loss of time
If high acceleration factors are used in undersampled k-space acquisition, then the scan time is reduced, but the diagnostic value of multi-contrast imaging is compromised
Solution Approach 1:
The patent merges multiple contrast images with different tissue weightings into a unified reconstruction framework. By combining T1-weighted, T2-weighted, and other contrast images, the system recovers fine anatomical details and tissue characteristics that would be lost in individual images, thereby maintaining diagnostic quality even with high acceleration factors.
Solution Approach 2:
The patent adds the contrast dimension to the reconstruction problem, transforming a 2D spatial reconstruction into a multi-dimensional problem that includes contrast weighting as an additional dimension. This allows the system to exploit redundancies across contrasts to recover high-quality images at high acceleration factors.
Data Source
AI summary
A plurality of sets of k-space data each of the same image region of a subject but having different contrasts are obtained. A sparse image coding procedure is performed to reconstruct a plurality of MR images each corresponding to one of the sets of k-space data. This involves solving an optimization problem comprising a data consistency iteration step used to generate the reconstructed MR images; and a denoising iteration step applied to the reconstructed MR images generated during the data consistency iteration step. The denoising iteration step includes performing a 2D/3D block matching operation to identify similar patches across the reconstructed MR images, and using the similar patches across the reconstructed MR images in a sparsifying operation to provide sparse representations of the reconstructed MR images. The sparse representations are used as an input to the data consistency iteration step.