Computational MRI with Physics-based Constraints: Application to Multi-contrast and Quantitative Imaging.

Area of Science:

• Medical Imaging
• Signal Processing
• Biophysics

Background:

• Compressed sensing (CS) in MRI utilizes implicit priors like sparsity for accelerated data acquisition.
• Traditional CS priors are phenomenological and do not incorporate underlying physics of image formation.
• Magnetic resonance signal dynamics are governed by physical laws that can be modeled explicitly.

Purpose of the Study:

• To introduce and explore physics-based modeling constraints in MRI reconstruction.
• To demonstrate the synergistic combination of explicit physical models and implicit priors within an inverse problem framework.
• To enable recovery of quantitative, bio-physical parameters alongside accelerated image reconstruction.

Main Methods:

• Incorporating physics-based signal models as explicit priors in the MRI reconstruction process.
• Combining these explicit priors with traditional implicit priors (e.g., sparsity) in a compressed sensing framework.
• Developing model-based quantitative MRI (Q-MRI) techniques and their linear subspace approximations.

Main Results:

• Achieved recovery of sharp, multi-contrast images from highly accelerated magnetic resonance imaging (MRI) scans.
• Demonstrated the ability to recover quantitative, bio-physical parameters by leveraging physics-based constraints.
• Presented successful MRI applications for multi-contrast imaging and quantitative mapping using the proposed framework.

Conclusions:

• Physics-based modeling offers a powerful approach to enhance compressed sensing in MRI.
• Explicit physical constraints improve image reconstruction quality and enable accurate quantitative parameter estimation.
• This framework facilitates advanced applications in multi-contrast imaging and quantitative MRI, improving diagnostic capabilities.