Variational multi-task MRI reconstruction: Joint reconstruction, registration and super-resolution

Area of Science:

• Medical Imaging
• Image Reconstruction
• Computational Biology

Background:

• Motion artifacts are a significant challenge in Magnetic Resonance Imaging (MRI), degrading image quality and hindering accurate diagnosis.
• Reconstructing high-quality, super-resolved images from undersampled MRI data corrupted by motion remains a critical problem.
• Existing methods often address reconstruction, registration, and super-resolution sequentially or in bi-task approaches, limiting overall performance.

Purpose of the Study:

• To develop a unified variational multi-task framework for simultaneous MRI reconstruction, registration, and super-resolution.
• To address motion degradation and undersampling in MRI data acquisition.
• To achieve superior motion-free, super-resolved image reconstructions compared to existing methods.

Main Methods:

• A novel variational multi-task optimization framework integrating reconstruction, registration, and super-resolution.
• Utilizing an L² fidelity term for shared representation, super-resolution foundations, and hyperelastic deformations for biological tissue modeling.
• Processing multiple undersampled MR acquisitions corrupted by motion within a single optimization model.

Main Results:

• Significant improvements in image quality, detail, and motion compensation compared to sequential and bi-task methods.
• Generation of sharp, highly textured images that preserve fine details.
• Demonstrated low computational time (CPU) alongside enhanced reconstruction quality.

Conclusions:

• The proposed variational multi-task framework effectively integrates MRI reconstruction, registration, and super-resolution for superior motion-free imaging.
• This unified approach outperforms state-of-the-art methods in producing high-quality, super-resolved MR images.
• The framework offers a promising solution for improving diagnostic accuracy in clinical MRI through enhanced image quality and reduced artifacts.