Reconstruction of dynamic image series from undersampled MRI data using data-driven model consistency condition...

Area of Science:

• Medical Imaging
• Magnetic Resonance Imaging (MRI)
• Image Reconstruction

Background:

• Dynamic MRI requires significant data, limiting imaging speed.
• Existing reconstruction methods often rely on low-rank temporal models, which can constrain solutions.
• Novel techniques are needed to accelerate MRI without compromising image quality.

Purpose of the Study:

• To develop a novel image reconstruction technique for accelerating dynamic MRI.
• To utilize pre-estimated low-rank temporal signal models from training data.
• To introduce the Model Consistency Condition (MOCCO) technique.

Main Methods:

• MOCCO regularizes reconstruction using temporal models without enforcing low-rank solutions.
• A data-driven model designs a transform for compressed sensing-type regularization.
• The method was compared against a standard low-rank approach in phantoms and patient data (CE-MRA, cardiac CINE).

Main Results:

• MOCCO showed reduced sensitivity to modeling errors compared to standard methods.
• Full-rank MOCCO solutions preserved temporal fidelity better.
• Significant aliasing and noise suppression were achieved in highly accelerated CE-MRA and cardiac CINE.

Conclusions:

• MOCCO overcomes deficiencies of previous temporal model-based methods.
• It enables high-quality image restoration from highly undersampled dynamic MRI data.
• The technique facilitates accelerated imaging in applications like CE-MRA and cardiac CINE.