Accelerated Imaging of Metallic Implants Using Model-Based Nonlinear Reconstruction

Total joint replacement is a common surgery to treat end-stage joint pain, which is projected to exceed 4.4 million cases by 2030 in the U.S. The severe artifacts near the metallic implant adopted in the surgery has made MRI unavailable for postoperative monitoring despite its high potential for the early detection of soft-tissue complications. Recent developments of multi-spectral imaging techniques demonstrated substantial improvements in the metal-induced artifact correction, but at the expense of elongated image acquisition time. In this work, we developed an image reconstruction algorithm using a signal model that describes the image distortion near the metallic implants. The incorporation of the signal model and the nonlinear reconstruction method exploiting it enabled high-quality image reconstruction of subsampled data compared to the straightforward application of generic methods for subsampled data for the accelerated imaging (parallel imaging and compressed sensing).

Shi X, Levine E, Weber H, Hargreaves BA. Accelerated imaging of metallic implants using model-based nonlinear reconstruction. Magn Reson Med. 2019 Apr;81(4):2247-2263.

Online Journal Article

Our proposed model-based reconstruction method generated a much sharper image than the conventional PI & CS method on the 2.6 times subsampled data compared to the reference (6-minute scan -> 2.3-minute scan). 

Professor of Radiology (Radiological Sciences Laboratory) and, by courtesy, of Electrical Engineering and of Bioengineering

Xinwei Shi, Evan Levine, and Hans Weber are alumni of the BMR group