Synthesizing Filter for Automatic MRI Image Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current parallel imaging techniques in MRI require user intervention for parameter settings, making them inconvenient and not fully automated.
Innovation Solution
A method and system for generating magnetic resonance (MR) images using a synthesizing filter, which is created based on calibration data sets to automatically reconstruct images from undersampled k-space data, utilizing convolution kernels adapted for both Cartesian and non-Cartesian sampling patterns.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If parallel imaging techniques are applied to accelerate MRI signal acquisition, then productivity is improved, but device complexity increases due to the need for multiple receiver coils and user intervention for parameter settings
Solution Approach 1:
The system performs self-calibration by automatically generating synthesizing filters from calibration data sets without requiring user intervention. The calibration process is executed autonomously by the processor, which determines calibration regions, constructs relationships between data points, and generates the necessary filtering parameters independently, thereby reducing operational complexity while maintaining accelerated imaging capabilities
Solution Approach 2:
A calibration data set is acquired and processed in advance to generate synthesizing filters before the actual imaging process. This preliminary calibration step stores electronic form synthesizing filters that can be directly applied during undersampled k-space data reconstruction, eliminating the need for real-time parameter adjustments and reducing system complexity during operation
2Manufacturing precision
If user intervention is required for parameter settings in parallel imaging, then manufacturing precision is maintained through manual control, but ease of operation deteriorates
Solution Approach 1:
The system automatically determines calibration regions and constructs data point relationships without user input. The processor independently executes the calibration algorithm, selecting calibration regions from the calibration data set and generating synthesizing filters autonomously, thereby maintaining precision through algorithmic accuracy while eliminating the need for manual parameter setting
Solution Approach 2:
The system uses calibration data sets containing fully-acquired k-space data to generate synthesizing filters that are then applied to undersampled data. This feedback mechanism ensures accurate reconstruction by comparing calibration-based filter performance against actual imaging data, maintaining precision while automating the process
3Manufacturing precision
If synthesizing filters are generated from calibration data sets, then manufacturing precision is improved through automatic reconstruction, but loss of time increases due to the calibration process
Solution Approach 1:
The calibration data set is acquired and synthesizing filters are generated in advance before actual imaging. This preliminary processing stores the filtering parameters electronically, allowing rapid application during reconstruction without repeating the full calibration process, thereby maintaining high reconstruction accuracy while reducing time loss during operational imaging
Solution Approach 2:
The calibration process focuses on determining specific calibration regions within the k-space data rather than processing the entire data set. By identifying and processing only the necessary calibration regions to generate synthesizing filters, the system achieves sufficient reconstruction accuracy with reduced processing time compared to full data processing
Data Source
AI summary
The disclosure relates to a system and method for generating or using a synthesizing filter in image reconstruction. The method may include: acquiring a calibration data set including a plurality of data points, determining a first calibration region in the calibration data set, the first calibration region including a matrix having a plurality of data points, the plurality of data points includes a first data point at the center of the first calibration region, constructing a first relationship between the first data point and the data points in the first calibration region, and generating a synthesizing filter based on the first relationship. The first data point is at the center of the first calibration region. The method may be implemented on at least one machine each of which has at least one processor and storage. The generated synthesizing filter may be stored in the storage in electronic form as a data file. The synthesizing filter may be adapted for determining an unknown data point in an undersampled k-space data set based on a signal acquired by the receiver coil.