Parallel MRI Reconstruction Using Low-Frequency Filter Banks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing parallel magnetic resonance imaging (pMRI) methods, such as SMASH, VD-AUTO-SMASH, GRAPPA, and 2D-GRAPPA, face challenges in image distortion compensation and require long reconstruction times, especially when dealing with multiple channels, due to the need for sensitivity profiles and separate channel reconstructions.
Innovation Solution
A method for parallel image reconstruction that involves acquiring image information via parallel coils, extracting low-frequency signals, generating filter banks for magnitude and phase information reconstruction, and combining these to produce a final image without requiring a sensitivity map, thereby reducing reconstruction time and compensating for image distortion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If separate channel reconstructions are performed using sensitivity profiles, then image information can be reconstructed, but reconstruction time increases significantly and image distortion occurs
Solution Approach 1:
The patent merges the reconstruction process by combining multiple channel images into a single reconstructed image using filter banks, eliminating the need for separate channel reconstructions. This integration reduces computational complexity and reconstruction time while maintaining image quality through the use of low-frequency k-space signal processing.
Solution Approach 2:
The patent extracts and utilizes only the low-frequency part of k-space signals for reconstruction, separating this essential component from the full k-space data. By focusing on low-frequency information which contains the majority of image content, the method reduces processing requirements and acceleration factors while maintaining reconstruction accuracy.
2Measurement precision
If sensitivity profiles are used for channel combination, then image reconstruction is possible, but image distortion due to interference between channels cannot be compensated
Solution Approach 1:
The patent employs feedback mechanisms through the filter bank processing that automatically compensates for channel interference and distortion. The filter banks adaptively process the combined channel signals to correct for harmful interference effects, eliminating the need for manual sensitivity profile calibration and providing automatic distortion compensation.
Solution Approach 2:
The patent changes the processing parameters by operating in the low-frequency k-space domain rather than using sensitivity profiles in the image domain. This parameter transformation allows for natural compensation of channel interference effects through the filter bank processing, improving image quality without requiring explicit sensitivity profile information.
3Measurement precision
If multiple ACS information pieces are acquired for constant estimation, then constant accuracy improves, but image acquisition time increases
Solution Approach 1:
The patent applies partial action by acquiring and processing only the low-frequency portion of k-space signals rather than the entire k-space data. This partial processing approach provides sufficient information for accurate constant estimation and image reconstruction while significantly reducing the acquisition time required compared to full k-space sampling.
Solution Approach 2:
The patent performs preliminary processing by extracting low-frequency signals from the k-space data before the main reconstruction process. This preliminary action prepares the data in an optimized form that accelerates subsequent reconstruction operations and reduces the overall acquisition time while maintaining reconstruction accuracy.
Data Source
AI summary
A method for a parallel image reconstruction is disclosed. The method includes (a) acquiring image information by channel via parallel coils in a magnetic resonance imaging (MRI) scanner; (b) extracting low-frequency signals from the image information; (c) reconstructing low-frequency images from the low-frequency signals; (d) generating filter banks by using the low-frequency images; and (f) reconstructing a final image by using the filter banks. The generating of the filter banks includes separately generating low-frequency image information for reconstruction of magnitude information and low-frequency image information for reconstruction of phase information, and then separately generating a filter for reconstruction of the magnitude information and a filter for reconstruction of the phase information.