Accuracy of MRI vs CT imaging with particular reference to patient specific templates for total knee replacement surgery.

BACKGROUND: The present study compares the accuracy of MRI and CT imaging for the manufacture of patient-specific templates for total knee replacement surgery. METHODS: A total of 10 ovine knees were imaged using MRI and CT scanners. Each set of images was reconstructed in 3D and then used to manufacture physical models of each bone, using rapid prototyping technology. After imaging the soft tissues were removed and specific measurements of the bony anatomy compared with measurements from the MRI and CT models. RESULTS: Bone models generated from MRI scans were dimensionally less accurate than those generated from CT scans. Furthermore, the bone models generated from MRI scans were visibly inferior to those generated from the CT scans. CONCLUSIONS: Current MRI scans do not offer a viable alternative to CT. Although adaptation of the template system to accommodate MRI imaging is possible, the changes required are neither practical nor desirable.

Computer-assisted total knee arthroplasty using patient-specific templating.

Current techniques used for total knee arthroplasty rely on conventional instrumentation that violates the intramedullary canals. Accuracy of the instrumentation is questionable, and assembly and disposal of the numerous pieces is time consuming. Navigation techniques are more accurate, but their broad application is limited by cost and complexity. We aimed to prove a new concept of computer-assisted preoperative planning to provide patient-specific templates that can replace conventional instruments. Computed tomography-based planning was used to design two virtual templates. Using rapid prototyping technology, virtual templates were transferred into physical templates (cutting blocks) with surfaces that matched the distal femur and proximal tibia. We performed 45 total knee arthroplasties on 16 cadaveric and 29 plastic knees, including a comparative trial against conventional instrumentations. All operations were performed using patient-specific templates with no conventional instrumentations, intramedullary perforation, tracking, or registration. The mean time for bone cutting was 9 minutes with a surgical assistant and 11 minutes without an assistant. Computer-assisted analyses of six random computed tomography scans showed mean errors for alignment and bone resection within 1.7 degrees and 0.8 mm (maximum, 2.3 degrees and 1.2 mm, respectively). Patient-specific templates are a practical alternative to conventional instrumentations, but additional clinical validation is required before clinical use.