Gradient nonlinearity calibration and correction for a compact, asymmetric magnetic resonance imaging gradient system

Area of Science:

• Medical Imaging
• Magnetic Resonance Imaging Physics
• Image Reconstruction

Background:

• Magnetic Resonance Imaging (MRI) relies on linear spatial encoding gradients.
• Engineering limitations introduce gradient nonlinearity (GNL), causing geometric image distortion.
• Accurate GNL modeling is crucial for correcting image distortions.

Purpose of the Study:

• To characterize and correct GNL in a novel asymmetric MRI gradient system.
• To evaluate the necessity of higher-order and even-order terms for GNL modeling.
• To compare calibration-based GNL coefficients with simulation-based ones.

Main Methods:

• An iterative calibration method using a fiducial phantom was employed.
• Phantom scans were performed across the gradient's spherical volume.
• Spherical harmonic polynomial models up to 10th-order (odd/even and odd-only) were tested.

Main Results:

• Model coefficients for the asymmetric gradient system were successfully estimated.
• GNL correction using 10th-order coefficients reduced residual error to 0.36 mm.
• Even-order terms were essential for accurate GNL modeling; calibrated coefficients outperformed simulation-based ones.

Conclusions:

• Accurate GNL characterization of asymmetric gradient systems requires higher-order and even-order terms.
• The iterative calibration method effectively corrects geometric distortions in MRI.
• This approach achieves spatial accuracy comparable to conventional MRI systems.