Motion-Robust Multi-Shot DWI Reconstruction via Locally Low-Rank Matrices
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Multi-shot diffusion-weighted MRI reconstruction is limited by motion-induced phase mismatches, leading to image blurring and distortion, and existing methods either increase scan time or struggle with phase estimation, especially with high-frequency phase variations.
Innovation Solution
A motion-robust reconstruction method that bypasses phase estimation by using locally low-rank spatial-shot matrices, constructed from k-space segments, to iteratively reconstruct high-resolution diffusion-weighted images, reducing distortions and acquisition time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If multi-shot EPI is used to provide high-resolution diffusion-weighted images with reduced distortion, then image resolution and distortion reduction are improved, but significant aliasing artifacts and signal cancellation occur due to motion-induced phase mismatch
Solution Approach 1:
The patent segments the k-space data into multiple shots and constructs separate low-rank matrices for each shot. By dividing the reconstruction problem into local spatial regions and applying low-rank constraints to each segment, the method handles motion-induced phase variations in each segment independently, thereby reducing aliasing artifacts while maintaining high resolution
Solution Approach 2:
The patent changes the mathematical formulation by imposing low-rank constraints on the spatial-shot matrices rather than attempting direct phase estimation. This parameter change transforms the non-convex phase estimation problem into a convex optimization problem that can be solved reliably, improving image quality without sacrificing resolution
2Measurement precision
If self-navigators or extra-navigators are used to estimate the phase of each individual shot, then phase estimation accuracy is improved, but scan time increases
Solution Approach 1:
The patent makes the main diffusion-weighted imaging data serve dual purposes: both for image reconstruction and for implicit phase estimation. By using the acquired DWI data itself to construct low-rank matrices and estimate phases through the convex optimization framework, the method eliminates the need for separate navigator acquisitions, thereby maintaining phase estimation accuracy without increasing scan time
Solution Approach 2:
The patent makes the acquired k-space data universally serve multiple functions: it is used simultaneously for image reconstruction, phase estimation, and motion correction. This multi-functionality approach replaces dedicated navigator echoes with the primary imaging data itself, reducing total acquisition time while maintaining measurement precision
3Measurement precision
If parallel imaging is used to reconstruct each shot separately with low-resolution results for phase estimation, then phase estimation is achieved, but image resolution is reduced and performance depends on array coil geometry
Solution Approach 1:
The patent segments both the spatial domain and shot domain to create local low-rank matrices. By dividing the image space into local regions and forming matrices that combine spatial information across shots within each region, the method achieves accurate phase estimation without requiring low-resolution preprocessing, thereby maintaining high image resolution
Solution Approach 2:
The patent adds a new dimension to the reconstruction problem by creating spatial-shot matrices that combine spatial information across multiple shots. This dimensional transformation allows simultaneous exploitation of spatial correlations and shot-to-shot relationships, achieving accurate phase estimation and high resolution without relying on array coil geometry
4Measurement precision
If phase estimation methods are used to handle motion-induced phase variations, then reconstruction accuracy is improved, but computational complexity increases and phase estimation may fail with high-frequency phase variations
Solution Approach 1:
The patent fundamentally changes the problem formulation by imposing low-rank constraints on spatial-shot matrices, which transforms the non-convex phase estimation problem into a convex optimization problem. This parameter change ensures reliable convergence and reduces computational complexity while handling high-frequency phase variations through the inherent flexibility of the low-rank model
Data Source
AI summary
Multi-shot diffusion-weighted magnetic resonance imaging acquires multiple k-space segments of diffusion-weighted MRI data, estimates reconstructed multi-shot diffusion weighted images, and combines the estimated images to obtain a final reconstructed MRI image. The estimation of images iteratively calculates updated multi-shot images from the multiple k-space segments and current multi-shot images using a convex model without estimating motion-induced phase, constructs multiple locally low-rank spatial-shot matrices from the updated multi-shot images, and calculates current multi-shot images from spatial-shot matrices.