MRI Parallel Imaging Reconstruction via k-Space Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Magnetic Resonance Imaging (MRI) techniques face challenges in achieving high-speed imaging due to reconstruction noise, especially with large reduction factors in parallel imaging methods like SENSE and k-t SENSE, which can damage the reproducibility of detailed structures in MR images.
Innovation Solution
The proposed solution involves a magnetic resonance imaging apparatus with a collector, transformation, unfolding, and inverse transformation modules that eliminate signal points based on a specific criterion to reduce reconstruction error and improve image quality, particularly by applying Fourier transforms and sensitivity distribution information to reduce the g-factor and noise amplification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If parallel imaging with large reduction factor is used to achieve high-speed imaging, then imaging speed is improved, but reconstruction noise increases and image quality deteriorates
Solution Approach 1:
The patent segments the k-space data into multiple segments and processes each segment separately through the unfolding operation. This segmentation allows for more controlled parallel imaging reconstruction, reducing the propagation of reconstruction noise while maintaining the speed benefits of parallel imaging with large reduction factors
Solution Approach 2:
The patent introduces a new dimension in the unfolding process by applying the operation not only in the spatial domain but also in the transform domain (after Fourier transform). This dimensional extension provides additional degrees of freedom for noise suppression while preserving image quality, effectively resolving the contradiction between imaging speed and image quality
2Manufacturing precision
If regularization techniques are used to reduce reconstruction noise, then noise is reduced, but dependency on prior knowledge increases and detailed structure reproducibility is damaged
Solution Approach 1:
The patent employs an unfolding operation that is self-adaptive to the acquired k-space data without requiring external prior knowledge or regularization parameters. The method automatically exploits the inherent structure in the parallel imaging data to suppress noise while preserving detailed structures, making the system self-sufficient and avoiding information loss associated with regularization
3Manufacturing precision
If full sampling is used to maintain image quality, then image quality is preserved, but imaging time increases and imaging speed is reduced
Solution Approach 1:
The patent performs preliminary sensitivity calibration and establishes the unfolding operation framework in advance. This preliminary action enables the use of accelerated sampling patterns with large reduction factors while maintaining image quality, as the unfolding operation is already prepared to correctly reconstruct the images from the reduced data without requiring full sampling
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enhances image quality by reducing reconstruction error and noise amplification, allowing for higher quality MR images even with large reduction factors, thereby improving the reproducibility of detailed structures without relying heavily on prior knowledge.
Implementation Method 1
Magnetic resonance imaging (MRI) apparatus is apparatus that visualizes internal body information of a subject utilizing a nuclear magnetic resonance phenomenon
Implementation Method 2
The transformation module obtains transformed space data of the respective channels by applying, to the time-series k-space data of the respect channels, Fourier transform on a spatial axis and certain transformation on a temporal axis
Data Source
AI summary
A magnetic resonance imaging apparatus according to an embodiment includes a collector, a transformation module, an unfolding module and an inverse transformation module. The collector collects time-series k-space data of a plurality of channels while spatially changing a sampling position. The transformation module obtains transformed space data of the respective channels by applying, to the time-series k-space data of the respect channels, Fourier transform on a spatial axis and certain transformation on a temporal axis. The unfolding module eliminates a signal point on a basis of a certain criterion and performs unfolding using the transformed space data on the respective channels and sensitivity distribution information on the respective channels; and the inverse transformation module applies inverse transformation of the certain transformation on the temporal axis to an unfolded data on which the signal point has been eliminated and the unfolding has been performed.