A variable splitting based algorithm for fast multi-coil blind compressed sensing MRI reconstruction.

Area of Science:

• Medical Imaging
• Signal Processing
• Optimization Algorithms

Background:

• Blind compressed sensing (BCS) leverages data-driven dictionaries for improved performance over fixed-dictionary compressed sensing (CS).
• A major limitation of BCS is its high computational complexity, hindering its application in Magnetic Resonance Imaging (MRI).

Purpose of the Study:

• To develop a novel optimization algorithm for BCS that enhances convergence speed.
• To address the computational challenges associated with BCS in MRI applications.

Main Methods:

• Proposed a new optimization algorithm utilizing variable splitting strategies for BCS.
• Decoupled sparse coefficient and dictionary update steps from the data fidelity term.
• Achieved closed-form analytical solutions for subproblems, avoiding slower iterative methods.

Main Results:

• Demonstrated superior performance of BCS over conventional CS schemes on multi-coil parametric MRI data.
• Achieved convergence speed-up factors exceeding 10-fold compared to previous BCS implementations.
• Validated the practical utility of BCS through experimental results.

Conclusions:

• The novel BCS optimization algorithm significantly improves convergence speed.
• This advancement makes BCS a more viable and efficient technique for MRI applications.
• The proposed method offers a substantial computational advantage over existing BCS algorithms.