Magnetic Resonance k-Space Readout via Random Sampling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current magnetic resonance imaging techniques face challenges in optimizing the time required to read out k-space data while preventing artifacts in the reconstructed image, particularly due to the limitations of the Nyquist condition.
Innovation Solution
The method involves using nonlinear gradients and violating the Nyquist condition by reading out k-space with a random pattern at reduced density, which allows for accelerated data acquisition and minimizes image artifacts through the principles of compressed sensing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If k-space is read out according to the Nyquist condition with sufficient density, then all signal variations can be detected and image artifacts are prevented, but the acquisition time becomes excessively long
Solution Approach 1:
The patent applies partial action by reading out only a subset of k-space points (randomly selected points) rather than all points required by the Nyquist condition. This partial readout reduces acquisition time while the compressed sensing algorithm later reconstructs the complete image, achieving a trade-off between speed and quality.
Solution Approach 2:
The patent changes the readout pattern parameter from systematic (Nyquist-condition compliant) to random sampling. This parameter change allows violation of the Nyquist condition in the acquisition domain while maintaining image quality through the random sampling property that enables compressed sensing reconstruction.
2Productivity
If k-space is read out with reduced density to accelerate acquisition, then acquisition time is reduced, but image artifacts arise due to insufficient data
Solution Approach 1:
The patent replaces the traditional mechanical/data-driven approach (systematic k-space readout following Nyquist condition) with a probabilistic approach (random sampling). This substitution allows reduced readout density while the random sampling pattern provides sufficient information for compressed sensing reconstruction, avoiding artifacts.
Solution Approach 2:
The patent employs feedback through iterative compressed sensing algorithms that reconstruct the image from the undersampled k-space data. The algorithm continuously refines the reconstruction based on the random sampling pattern, ensuring image quality is maintained despite reduced readout density.
3Measurement precision
If systematic gradients are used for spatial coding, then spatial resolution is achieved, but the readout trajectory becomes constrained and cannot optimize acquisition time
Solution Approach 1:
The patent introduces dynamics by replacing fixed systematic gradient trajectories with dynamic random sampling patterns. The random selection of k-space points allows flexible optimization of readout timing and trajectory, enabling faster acquisition while maintaining spatial resolution through the random sampling property.
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 enables faster acquisition of magnetic resonance measurement data while avoiding correlations between read-out k-space points, allowing for artifact-free image reconstruction even with underdetermined systems.
Implementation Method 1
The deflection typically takes place by radiation of radio-frequency excitation pulses (RF excitation pulses) into the examination subject
Implementation Method 2
The gradient field varies the resonance frequency (Larmor frequency) and, for example, the phase position of the magnetization deflected by an RF pulse
Implementation Method 3
The magnitude of the magnetization (in particular of the transverse magnetization, defined in a plane transverse to the basic magnetic field) at a defined location of the examination subject can be determined from the readout point with the use of a Fourier transformation
Data Source
AI summary
In a method, a magnetic resonance system, and a computer-readable storage medium to generate magnetic resonance measurement data of an imaging area of an examination subject, the imaging area being located in a measurement volume of the magnetic resonance system, by operation of the magnetic resonance system, during the acquisition of the magnetic resonance measurement data, at least one additional nonlinear gradient is switched in addition to the gradients for spatial coding, and k-space is read out according to a random pattern, less densely than is required by the Nyquist condition.