MRI Reconstruction Combining Compressed Sensing and Parallel Imaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Combining parallel imaging with compressed sensing in MRI systems reduces computational efficiency, negating some benefits of individual techniques, such as reduced scan time and image quality.
Innovation Solution
A method that involves acquiring k-space data sets with calibration data in the center and randomly undersampled data in the outer region, applying compressed sensing reconstruction, followed by a Fourier transform, and then using parallel imaging reconstruction to synthesize unacquired data, ultimately generating a complete k-space data set for high-quality image reconstruction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If parallel imaging and compressed sensing are combined in MRI reconstruction, then scan time is reduced and image quality is improved, but computational efficiency deteriorates
Solution Approach 1:
The patent segments the k-space data into two distinct regions: a center region containing calibration data and an outer region containing randomly undersampled imaging data. This segmentation allows different reconstruction approaches to be applied to different regions, optimizing both image quality and computational efficiency while maintaining the benefits of combined parallel imaging and compressed sensing
Solution Approach 2:
The patent applies different sampling strategies to different regions of k-space: uniform sampling in the center region for calibration purposes and random undersampling in the outer region for accelerated imaging. This local differentiation enables the system to maintain high computational efficiency in the calibration region while achieving scan time reduction in the imaging region
2Manufacturing precision
If parallel imaging and compressed sensing are combined in MRI reconstruction, then image quality is improved, but computational complexity increases
Solution Approach 1:
The patent divides the reconstruction process into separate stages corresponding to different k-space regions. The center region calibration data is processed independently to determine coil sensitivity profiles, while the outer region undersampled data is processed using compressed sensing algorithms. This segmentation reduces overall computational complexity by avoiding the need to process the entire k-space with the most computationally intensive method
Solution Approach 2:
The patent extracts and removes calibration data from the center of k-space before applying compressed sensing reconstruction to the remaining imaging data. This extraction allows the calibration information to be utilized separately for determining reconstruction parameters, thereby reducing the computational burden on the main reconstruction algorithm while maintaining high image quality
Data Source
AI summary
A method for generating a magnetic resonance image includes acquiring a first k-space data set from each of a plurality of RF coils. The first k-space data set includes calibration data and randomly undersampled data. For each RF coil, a fully randomly sampled k-space data set is generated by removing a portion of the calibration data. A compressed sensing reconstruction technique is applied to the fully randomly sampled k-space data set to generate an aliased image, which is used to generate a uniformly undersampled k-space data set. A second k-space data set is generated by inserting the portion of the calibration data and a parallel imaging reconstruction technique is applied to the second k-space data set to synthesize unacquired data. The second k-space data set and the synthesized data are combined to generate a complete k-space data set for the RF coil.