SPARKLING k-space Trajectories for MRI Acceleration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Magnetic Resonance Imaging (MRI) procedures are lengthy, especially when acquiring large or high-resolution images, due to the need for extensive k-space sampling to meet the Nyquist criterion, leading to long acquisition times and potential aliasing artifacts.
Innovation Solution
The method employs non-parametric 'SPARKLING' trajectories for k-space sampling, which are optimized to increase sampling efficiency by oversampling in the central k-space region, allowing for reduced acquisition time without degrading image quality, using an algorithm that projects a target sampling distribution onto feasible trajectories constrained by hardware and physiological limits.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional k-space sampling schemes are used to meet the Nyquist criterion, then image quality is maintained, but acquisition time becomes excessively long
Solution Approach 1:
The patent applies partial action by sampling only a subset of k-space points rather than the full Nyquist-required grid. Compressed sensing theory enables reconstruction from this partial sampling by exploiting image sparsity in a transform domain, achieving acceptable image quality with significantly fewer measurements than conventional methods
Solution Approach 2:
The patent transforms the sampling problem from the spatial domain to a sparsity-promoting transform domain (such as wavelet or total variation domain). By changing the representation parameters and solving an optimization problem that enforces sparsity constraints, the system can reconstruct high-quality images from undersampled k-space data
2Loss of time
If multiple receiver coils are used for parallel MRI or Simultaneous Multislice imaging, then acquisition time is reduced, but hardware cost and complexity increase
Solution Approach 1:
The patent enables the single receiver coil to perform the function that would otherwise require multiple coils. By using compressed sensing with sparsity constraints, the system makes the redundant information inherent in the undersampled data sufficient for reconstruction, eliminating the need for additional hardware while achieving similar acceleration
3Loss of time
If k-space is under-sampled to accelerate acquisition, then acquisition time is reduced, but aliasing artifacts appear
Solution Approach 1:
The patent converts the harmful aliasing artifacts into useful information. By designing the undersampling pattern and sparsity constraint together, the aliasing patterns that would normally corrupt the image instead provide redundant equations that help solve for the true image in the sparsity-promoting domain, ultimately improving reconstruction accuracy
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 significantly reduces acquisition time while maintaining high image quality, achieving an acceleration factor of up to 10 times faster than conventional methods without compromising image fidelity, as demonstrated by improved SSIM scores and reduced NRMSE.
Implementation Method 1
Sampling is performed by acquiring the nuclear magnetic resonance (NMR) signal generated by excited nuclear spins at predetermined times, which correspond to points along said trajectory. The trajectory is defined by a time-varying magnetic field gradient applied to the body to be imaged after the excitation of its nuclear spins by a radio-frequency (RF) pulse.
Implementation Method 2
Sampling is performed by acquiring the nuclear magnetic resonance (NMR) signal generated by excited nuclear spins at predetermined times
Data Source
Figure 1~2
Figure 3~4B
Figure 5A~6B
AI summary
A method of performing magnetic resonance imaging of a body comprising: a. immerging the body in a static and substantially uniform magnetic field; b. exciting nuclear spins inside said body using at least one radio-frequency pulse; c. applying to said body a time-varying magnetic field gradient defining at least one trajectory (ST) in k-space and simultaneously acquiring samples of a magnetic resonance signal so as to perform a pseudo-random sampling (KS) of the k-space; and d. applying a sparsity-promoting nonlinear reconstruction algorithm for reconstructing a magnetic resonance image of said body; wherein, at least in a low-spatial frequency region of the k- space, the distance between any two adjacent points belonging to a same trajectory is lower than 1/FOV, FOV being the size of a field of view of the reconstructed image. A magnetic resonance imaging apparatus for carrying out such a method.