Jacobian-Based Correction of 3D-MSI Images Near Implanted Metal Devices

Kevin M. Koch¹, Matthew A. Koff², Hollis G. Potter²

3D-Multi-Spectral Imaging techniques (MAVRIC, SEMAC, and combinations thereof) can substantially reduce image artifacts in the presence of implanted metal. Residual susceptibility artifacts in 3D-MSI techniques are commonly manifested as localized pileup artifacts - where many spins are frequency encoded to the same spatial position. Here, we show that spectral information inherent to 3D-MSI acquisitions can be used to construct field maps with sufficient accuracy to combat these residual artifacts through estimation of local gradients and Jacobian intensity correction of shifted pixels. The proposed methods are demonstrated on clinical images at 1.5T of total knee and shoulder replacements composed of cobalt-chromium alloys.