Self-calibrated interpolation of non-Cartesian data with GRAPPA in parallel imaging

Area of Science:

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

Background:

• Non-Cartesian imaging allows for faster MRI acquisition but requires complex reconstruction.
• Through-time GRAPPA (TT-GRAPPA) improves reconstruction quality for accelerated non-Cartesian data using region-specific kernels.
• TT-GRAPPA necessitates extensive calibration scans, increasing overall imaging time.

Purpose of the Study:

• To develop a novel non-Cartesian k-space reconstruction method.
• To enable self-calibration of region-specific interpolation kernels for highly accelerated acquisitions.
• To reduce the calibration scan time in dynamic MRI.

Main Methods:

• Proposed Self-calibrated Interpolation of Non-Cartesian data with GRAPPA (SING) method.
• SING synthesizes calibration data from undersampled measurements to self-calibrate kernels.
• Employed local k-space regularization via an extension of TT-GRAPPA.
• Compared SING with conjugate-gradient SENSE and TT-GRAPPA using numerical phantoms and in vivo cine data.

Main Results:

• SING demonstrated visually and quantitatively comparable reconstruction quality to TT-GRAPPA.
• SING outperformed conjugate-gradient SENSE in reconstruction quality.
• Temporal fidelity of SING and TT-GRAPPA was similar at equivalent acceleration rates.
• SING and TT-GRAPPA exhibited similar noise amplification (G-factor) at moderate to high acceleration rates.

Conclusions:

• The proposed SING reconstruction significantly improves calibration data acquisition efficiency.
• SING matches the reconstruction performance of TT-GRAPPA.
• SING offers a promising alternative for accelerated non-Cartesian MRI, reducing scan time without compromising image quality.