MRI Image Reconstruction Using Multiple Sparse Spaces
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional magnetic resonance imaging (MRI) techniques face limitations in reducing imaging time due to the need for extensive data acquisition, despite the introduction of compressed sensing methods, which require optimal sparse spaces for effective signal reconstruction.
Innovation Solution
The method involves setting multiple sparse spaces based on image characteristics and implementing independent compressed sensing algorithms for each space, allowing for improved resolution with reduced data acquisition, utilizing techniques like total variation and wavelet transforms to separate and reconstruct image components.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If conventional compressed sensing technique is used with single sparse space, then imaging time is reduced, but image resolution and quality deteriorate for complicated structures
Solution Approach 1:
The patent divides the image reconstruction process into multiple segments by applying different sparse spaces to different regions or components of the image. The image is segmented into multiple parts, each processed with an optimized sparse space, thereby maintaining high resolution for complicated structures while preserving the time-saving benefits of compressed sensing.
Solution Approach 2:
The patent applies the principle of local quality by using different sparse spaces for different regions of the image based on their specific characteristics. Regions with complicated structures use sparse spaces optimized for those structures, while other regions use appropriate sparse spaces, thereby achieving locally optimized image quality without sacrificing overall imaging speed.
2Measurement precision
If multiple sparse spaces are applied for different image regions, then image resolution is improved, but processing complexity increases
Solution Approach 1:
The patent segments the image processing into multiple independent tasks, each handling a specific region or component with its own sparse space. This segmentation allows parallel processing of different image parts, which distributes the computational complexity across multiple simpler tasks rather than one complex task, thereby managing processing complexity more effectively.
Solution Approach 2:
The patent applies partial action by selectively using multiple sparse spaces only where needed (in regions with complicated structures) rather than applying all sparse spaces to the entire image. This selective approach maintains high resolution where necessary while avoiding the unnecessary computational complexity in regions where simpler processing suffices.
3Loss of time
If variable density random under sampling is used, then imaging time is reduced, but aliasing artifacts and noise increase
Solution Approach 1:
The patent converts the harmful aliasing artifacts and noise into beneficial information by using them as constraints in the reconstruction process. The multi-sparse space approach leverages the sparsity patterns in different transform domains to distinguish and remove aliasing artifacts, transforming what would be harmful interference into useful constraints that guide accurate image reconstruction.
Solution Approach 2:
The patent changes the parameters of the reconstruction process by using multiple sparse spaces with different properties. By transforming the image data into multiple different domains (e.g., wavelet, gradient, curvature spaces), the system can adjust the reconstruction parameters to optimize for removing aliasing artifacts while preserving true image features, thereby reducing harmful artifacts even with accelerated sampling.
Data Source
AI summary
A method for processing a magnetic resonance image is provided. The method includes receiving data through a receiving coil; setting a plurality of sparse spaces for unit data of a single image; reconstructing an image by data, for which the sparse spaces have been set; and combining the reconstructed images with each other to provide the combined image.