On-the-Fly Adaptive ${k}$ -Space Sampling for Linear MRI Reconstruction Using Moment-Based Spectral Analysis

Area of Science:

• Medical Imaging
• Applied Mathematics
• Signal Processing

Background:

• Accelerated magnetic resonance imaging (MRI) relies on k-space sampling strategies.
• Current methods often use image-domain metrics, which are computationally expensive.
• Designing optimal sampling patterns to mitigate noise amplification remains challenging.

Purpose of the Study:

• To develop a computationally efficient theoretical framework for designing MRI sampling patterns.
• To relate geometric properties of sampling patterns to noise amplification.
• To enable automatic and adaptive design of multidimensional Cartesian sampling patterns.

Main Methods:

• Introduced a theoretical framework based on local geometric properties of sampling patterns.
• Related these properties to the spectral moments of the information matrix.
• Utilized this criterion for efficient optimization of sampling patterns without image reconstruction.

Main Results:

• Demonstrated a novel criterion for evaluating sampling patterns.
• Showed that this criterion effectively mitigates noise amplification.
• Achieved strong agreement with traditional image-domain and k-space metrics in experiments.

Conclusions:

• The proposed theoretical framework offers a computationally efficient approach for designing MRI sampling patterns.
• This method allows for on-the-fly, automatic, and adaptive optimization.
• It has the potential to significantly improve the speed and efficiency of high-dimensional MRI acquisition.