抄録
Three-dimensional (3D) bone surface models reconstructed from computed tomography (CT) scan data have been used for surgical planning and intra-operative registration. Although CT scanning is an excellent technique for describing bone structure and geometry, it has disadvantages of radiation exposure and poor ability of visualizing soft tissues such as articular cartilage. Magnetic resonance (MR) is a powerful method for imaging soft tissues without radiation. Recently quantification of cortical bone structure and geometry has been investigated using MR imaging (MRI). The aim of the present study is to examine the precision of 3D MRI-based bone surface models. Porcine femora and tibiae were used for the study. Three polymeric spherical markers containing gadolinium contrast media were fixed to the femur and the tibia, respectively. MRI and CT scanning were performed with routine protocols. 3D bone surface models were reconstructed with a commercial software package. Actual bone dimensions were measured with a caliper at the cross sections where the markers were fixed. These dimensions were compared with those on MRI-based and CT-based models. The distances between two of the three markers were also compared. The maximum differences in dimension between actual bone and model were 2.6 mm for sagittal MRI, 1.4 mm for axial MRI and 0.8 mm for CT, respectively. Errors in the marker distance were within 1.0 mm for both MRI and CT models, showing no significant differences between both models. Automatic tracing of bone outlines may improve the accuracy of MRI models.
