Online Spatiotemporally Constrained Reconstruction for Real-Time Interactive MRI

Area of Science:

• Medical Imaging
• Magnetic Resonance Imaging
• Computational Imaging

Background:

• Dynamic MRI reconstruction is crucial for real-time imaging, but current online methods often lack sufficient spatial and temporal fidelity.
• Fetal cardiac imaging presents unique challenges due to rapid motion, requiring high-resolution and low-latency imaging techniques.

Purpose of the Study:

• To develop and evaluate a low-latency dynamic MRI reconstruction method, online spatiotemporally constrained reconstruction (STCR).
• To achieve superior spatial and temporal fidelity compared to existing online reconstruction techniques for improved localization, particularly in fetal cardiac imaging.

Main Methods:

• Implemented online spatiotemporally constrained reconstruction (STCR) utilizing an efficient augmented Lagrangian solution with a limited iteration count.
• Evaluated online STCR for fetal cardiac localization at 0.55 Tesla, comparing it against online view-sharing and offline compressed sensing using edge sharpness metrics.

Main Results:

• Online STCR demonstrated the highest structural similarity index measure (SSIM) to offline STCR among tested online methods.
• Achieved a 13% ± 2% improvement in spatial edge sharpness and a 34% ± 4% improvement in temporal edge sharpness compared to view-sharing.
• Online STCR successfully reconstructed 2D real-time MRI data with 1.5 x 1.5 mm spatial resolution and 31.8 ms temporal resolution.

Conclusions:

• Online STCR provides a low-latency dynamic MRI reconstruction solution suitable for real-time applications.
• The method significantly enhances the depiction of rapidly moving boundaries, outperforming standard view-sharing techniques.
• Demonstrated feasibility for high-resolution, high-temporal-fidelity fetal cardiac MRI localization.