[Compressed sensing magnetic resonance image reconstruction based on double sparse model].

Area of Science:

• Medical imaging
• Signal processing
• Computational science

Context:

• Magnetic resonance (MR) image reconstruction is vital for magnetic resonance imaging (MRI).
• Compressed sensing (CS) theory enables accurate reconstruction from undersampled measurements using image sparsity.
• Improving MR image reconstruction quality via enhanced sparse priors is a critical challenge.

Purpose:

• To propose an adaptive image reconstruction model for compressed sensing MRI.
• To exploit sparse priors in both image and transform domains using a double dictionary learning approach.
• To combine synthesis and sparse transform models for complementary sparse representations.

Summary:

• An adaptive model fuses synthesis and sparse transform models via double dictionary learning for CS MR image reconstruction.
• The model utilizes iterative alternating minimization and synthesis/transform K-singular value decomposition (K-SVD) algorithms.
• This approach leverages complementary sparse priors from both domains for improved reconstruction.

Impact:

• The proposed model demonstrates superior image reconstruction quality compared to existing MRI methods.
• It offers a faster convergence rate and enhanced robustness against noise.
• This advancement contributes to more efficient and reliable MRI data processing.