Quantifying image distortion of orthopedic materials in magnetic resonance imaging

Abstract

To determine the magnitude of image distortion between two-dimensional (2D) fast-spin-echo (FSE) images and 3D-MAVRIC by using a phantom with samples of common materials used in total joint arthroplasty. A phantom was constructed to hold samples of 316 stainless steel, cobalt chrome, titanium, and ultra-high molecular weight polyethylene (UHMWPE), and to permit tracking of points between 2D-FSE and 3D-MAVRIC sequences. Imaging was performed with a 1.5 Tesla scanner. The displacement of points between the two acquisitions in regions of varying distance from the material sample was calculated. Measured displacements were compared with theoretical displacements calculated from MAVRIC frequency field maps, and to the known phantom dimensions. Bulk in-plane artifacts increased from the control scan (0.20 ± 0.07 mm), to UHMWPE (0.23 ± 0.04 mm), titanium (0.40 ± 0.34 mm), cobalt chrome (1.35 ± 0.57 mm), and to stainless steel (2.56 ± 0.62 mm). A similar pattern was found for bulk through-plane measurements: control scan (0.36 ± 0.08 mm), UHMWPE (0.38 ± 0.05 mm), titanium (1.11 ± 0.51 mm), and cobalt chrome (2.08 ± 1.83 mm). Large distortions were observed near the metal samples, and reduced with distance from the samples. The differences between the measured displacement and theoretical displacement was typically less than one pixel or one slice dimension for in-plane and through-plane measurements, respectively. The difference between the 3D-MAVRIC and known dimensions of the phantom was less than two pixels, whereas the results for the 2D-FSE were less consistent. Distortion was reduced in 3D-MAVRIC scans. Measured distortions corresponded well to theoretical calculations from frequency field map data.