Coupled Dictionary Learning for Multi-Contrast MRI Reconstruction.

Area of Science:

• Medical Imaging
• Signal Processing
• Machine Learning

Background:

• Magnetic resonance imaging (MRI) often requires multiple contrasts (e.g., T1-weighted, T2-weighted, FLAIR) to capture comprehensive anatomical information.
• These different contrasts share underlying anatomical similarities, presenting an opportunity for joint analysis and reconstruction.
• Reconstruction from under-sampled k-space data is crucial for accelerating MRI acquisition but introduces aliasing and noise.

Purpose of the Study:

• To develop a novel multi-contrast MRI reconstruction method that exploits inter-contrast correlations.
• To improve the quality and efficiency of reconstructing multiple MRI contrasts from under-sampled k-space data.
• To demonstrate the advantages of the proposed method in capturing structural dependencies and its potential for quantitative MRI.

Main Methods:

• A coupled dictionary learning based multi-contrast MRI reconstruction (CDLMRI) approach is proposed.
• The method iterates through three stages: coupled dictionary learning, coupled sparse denoising, and enforcing k-space consistency.
• Dictionaries are learned to be adaptive to individual contrasts while capturing cross-contrast correlations in a sparse domain.

Main Results:

• The CDLMRI approach successfully leverages structural dependencies between different MRI contrasts.
• Numerical experiments with retrospective under-sampling demonstrate effective noise and aliasing removal.
• The learned priors show significant advantages for multi-contrast MRI reconstruction.

Conclusions:

• The proposed CDLMRI method effectively utilizes correlations between different MRI contrasts for guided or joint reconstruction.
• This approach offers improved reconstruction quality and holds promise for quantitative MRI applications like MR fingerprinting.
• The learned dictionaries provide powerful priors for enhancing multi-contrast MRI reconstruction.