Phantom based qualitative and quantitative evaluation of artifacts in MR images of metallic hip prostheses.

PURPOSE To develop methods for qualitative and quantitative evaluation of MRI artifacts near metallic prostheses, and to compare the efficiency of different artifact suppression techniques with different types of hip prostheses. METHODS Three hip prostheses of cobalt-chromium, stainless steel, and titanium were embedded in agarose gel together with a rectilinear grid. Coronal MR images of the prostheses were acquired on a 1.5T scanner. Three pulse sequences were evaluated; TSE: a high-bandwidth turbo spin echo; VAT: TSE with view angle tilting, SEMAC: TSE with both VAT and slice distortion correction (6, 10 or 16 z-phase-encoding steps). Through-plane distortions were assessed as the length of visible gridlines, in-plane artifacts as the artifact area, and total artifacts by subtraction of an ideal, undistorted image from the actual image. RESULTS VAT reduced in-plane artifacts by up to 50% compared to TSE, but did not reduce through-plane artifacts. SEMAC reduced through-plane artifacts by 60-80% compared to TSE and VAT. SEMAC in-plane artifacts were from 20% higher (6 encoding steps) to 50% lower (16 steps) than VAT. Total artifacts were reduced by 60-80% in the best sequence (SEMAC, 16 steps) compared to the worst (TSE). The titanium prosthesis produced 3-4 times lower artifact scores than the other prostheses. CONCLUSIONS A rectilinear grid phantom is useful for qualitative and quantitative evaluation of artifacts provoked by different MRI protocols and prosthesis models. VAT and SEMAC were superior to TSE with high bandwidth. A proper number of z-encoding steps in SEMAC was critical. The titanium prosthesis caused least artifacts.