Distortion Correction in TSE Near Metal Implants at 0.55T Using Model-Based Iterative Reconstruction and Opposite Readout Acquisitions

Motivation: View angle tilting for reducing signal displacement artifacts near metal is difficult to adopt at 0.55T due to blurring and decreased SNR due to high readout bandwidth. Goal(s): To develop a method that corrects in-plane displacement artifacts near metal implants at 0.55T, achieving higher SNR without sacrificing image sharpness. Approach: TSE acquisitions with opposite frequency-encoding directions were performed. Gradient nonlinearity-corrected images were first generated. Displacement maps were estimated by TOPUP from the GNL-corrected images. Final images were reconstructed using Type-1 NUFFT-based CG-SENSE by incorporating both GNL and TOPUP-derived displacement fields. Results: The proposed method, performed without VAT, eliminated blurring and achieved superior SNR. Impact: This method enhances SNR without blurring for MRI near metal implants at 0.55T, benefiting low-SNR regions such as the spine and disc, while avoiding the prolonged scan times associated with VAT methods that require multiple averages for SNR improvement.