Accelerating 4D flow MRI by exploiting low-rank matrix structure and hadamard sparsity

Area of Science:

• Magnetic Resonance Imaging
• Biomedical Engineering
• Cardiovascular Imaging

Background:

• Four-dimensional flow MRI (4D flow MRI) provides comprehensive hemodynamic information but suffers from long acquisition times.
• Accelerating 4D flow MRI is crucial for clinical applicability and patient comfort.
• Existing acceleration methods face limitations in reconstruction accuracy and speed.

Purpose of the Study:

• To develop an accelerated 4D flow MRI technique by exploiting the inherent low-rank structure of the data.
• To enhance reconstruction accuracy by incorporating Hadamard sparsity in the velocity-encoding segments.
• To validate the proposed method against current state-of-the-art reconstruction techniques.

Main Methods:

• Representing 4D flow MRI data as a sum of low-rank and sparse components.
• Applying a Hadamard transform to velocity-encoding segments for optimized sparse representation.
• Reconstructing undersampled aorta data from healthy subjects using the proposed method and comparing it with k-t SPARSE-SENSE.
• Demonstrating image reconstruction from eight-fold prospective undersampling and comparing with conventional SENSE.

Main Results:

• The proposed method demonstrated consistently lower velocity estimation errors compared to k-t SPARSE-SENSE in simulations.
• In vivo results showed reduced peak flow error in the ascending and descending aorta compared to k-t SPARSE-SENSE (e.g., 2.5% vs. 10.2% in ascending aorta).
• Streamline visualization revealed more consistent flow fields with the proposed technique.

Conclusions:

• Exploiting low-rank structure and Hadamard sparsity significantly improves 4D flow MRI reconstruction accuracy.
• The developed method offers a promising approach to reduce the long scan times associated with 4D flow MRI.
• This advancement has the potential to enhance the clinical utility of 4D flow MRI for cardiovascular assessments.