MRI Hybrid-Domain Reconstruction via Null-Subspace Bases
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current MRI image reconstruction methods, such as PRUNO and ESPIRIT, face challenges with noise amplification and artifacts due to inaccurate coil sensitivity information and manual thresholding, which are image-content dependent and computationally demanding, especially for dynamic imaging.
Innovation Solution
A hybrid-domain reconstruction method that extracts null-subspace bases from k-space calibration data to derive spatial nulling maps, eliminating the need for coil sensitivity masking and being insensitive to subspace division, thereby providing a robust and efficient reconstruction process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If ESPIRIT uses manual thresholding of eigenvalues to mask eigenvector maps, then noise propagation is minimized, but the process becomes computationally demanding and image-content dependent
Solution Approach 1:
The system performs self-calibration by automatically determining the subspace cut-off value through the ratio of signal-subspace dimension to total subspace dimension, without requiring manual intervention. The calibration matrix is constructed from the data itself, and the optimal cut-off is derived self-consistently from the singular value decomposition results
Solution Approach 2:
The invention changes the approach from manual eigenvalue thresholding to an automatic subspace cut-off method based on the ratio of signal-subspace dimension to total subspace dimension. This parameter change eliminates the need for trial-and-error threshold selection and reduces computational burden while maintaining reconstruction quality
2Ease of operation
If SURE-based ESPIRIT automatically selects parameters, then manual optimization is avoided, but computational time increases significantly for dynamic imaging
Solution Approach 1:
The invention uses a simplified criterion (ratio of subspace dimensions) rather than exhaustive SURE optimization. This partial action approach provides sufficient accuracy for dynamic imaging without the excessive computational cost of full SURE-based parameter selection
Solution Approach 2:
The subspace cut-off value is determined preliminarily from the calibration data before the actual dynamic image reconstruction. This preliminary determination using the subspace dimension ratio avoids repeated computational optimization during the reconstruction process
3Reliability
If PRUNO uses null-subspace bases for k-space nulling, then residual artifacts are reduced, but the method is computationally prohibitive and cannot incorporate image priors
Solution Approach 1:
The invention segments the problem into two parts: (1) using null-subspace bases from calibration data to form the reconstruction matrix, and (2) applying this matrix to dynamic image data. This segmentation allows pre-computation of the nulling operation from calibration data, making the actual reconstruction computationally efficient
Solution Approach 2:
The reconstruction matrix derived from null-subspace bases serves multiple functions: it provides artifact reduction, enables fast reconstruction, and can be applied to multiple dynamic images. The universal applicability of the pre-computed matrix eliminates the need for repeated heavy computations
Data Source
AI summary
A method for hybrid-domain reconstruction of MRI images includes the steps of (A) extracting null-subspace bases of a calibration matrix from k-space coil calibration data to calculate image-domain spatial null maps (SNMs) and (B) reconstructing multi-channel images by solving an image-domain nulling system formed by SNMs that contain both coil sensitivity and finite image support information, thus circumventing the masking-related procedure and demonstrating a robust reconstruction.