On optimality of parallel MRI reconstruction in k-space

Area of Science:

• Medical Imaging
• Magnetic Resonance Imaging (MRI)
• Signal Processing

Background:

• Parallel MRI reconstruction in k-space offers advantages like error tolerance and efficient non-Cartesian data processing.
• These benefits stem from approximating unsampled k-space data using local samples.

Purpose of the Study:

• To optimize reconstruction kernels for parallel MRI in k-space.
• To evaluate the impact of regularization on image error.
• To assess the accuracy of k-space-based parallel MRI methods.

Main Methods:

• Reconstruction is framed as approximating the pseudoinverse with a sparse matrix.
• Reconstruction kernels are optimized using approximation error as a criterion.
• An algorithm for automatic kernel selection and a total error metric for validation are introduced.

Main Results:

• The novel methods demonstrate reduced reconstruction and noise errors compared to existing techniques.
• Validation was performed using both simulated and real MRI data.
• Optimized kernels and regularization parameters led to improved image quality.

Conclusions:

• The developed methods enhance parallel MRI reconstruction accuracy and efficiency.
• These techniques are valuable for minimizing image artifacts and accelerating data processing.
• The approach aids in validating parallel MRI reconstruction designs for specific coil systems and k-space trajectories.