Recently, as the number of total knee arthroplasty (TKA) is increasing, the number of revision TKA due to loosening or osteolysis is rapidly increasing. Large bone defect is one of the most critical issues during revision TKA. Therefore, early detection of bone loss around the TKA prosthesis before bone loss has been enlarged is very important. However, it is difficult to detect the loosening or ostolysis at the early stage around the femoral component even using fluoroscopically guided plain radiograph. A novel technique of tomography (Tomosynthesis; Shimazu Corporation, Kyoto, Japan) was introduced to detect the small bone loss. The purpose of this study was to examine, in a pig model of radiolucent line and osteolysis around TKA, the sensitivity and specificity of detection of radiolucent line and osteolysis using fluoroscopically guided plain radiographs and a novel technique of tomography. Six cemented femoral components (PFC Sigma; DePuy, Warsaw, IN, USA) were implanted in pig knees. Two components were implanted with standard cement technique (Standard model). Two components were implanted with 2 mm-thick defect between the cement and bone (Radiolucent line model). Two components were implanted with cystic defects (mean size = 0.7 cm3) in femoral condyles (Osteolysis model). The simulated bone lesions were filled with agarose to simulate granuloma tissue and to reduce the air artifact around the bone lesions, which can interfere with imaging techniques (Figure 1). Fluoroscopically guided plain radiographs (63 kV, 360 mA, 50 msec) were taken in 4 postures (antero-posterior, lateral, and +/−45 degrees oblique views) for each specimen (Figure 2). For Tomosynthesis, 74 frames were acquired in the rate 30 frames/sec with fixed X-ray condition (65 kV 1.25 mAs) and were reconstructed (Figure 3). Seven blinded assessors experienced in clinical radiographic analysis examined. The sensitivities, specificities and accuracy of the two imaging techniques were compared.INTRODUCTION
METHODS
In previous congress of ISTA in Hawaii, we reported the results about accuracy of the cup center position in our image-free navigation system. In the new version of our navigation system, leg elongation and offset change as well as cup center position can be navigated. In this study, we therefore investigated the accuracy of cup center position, leg elongation and offset change. Twenty four THA operations were performed with using the image-free OrthoPilot THA3.1 dysplasia navigation system (B. Braun Aesculap, Tuttlingen, Germany) between August 2009 and December 2009 by three experienced surgeons. In this system, cup center height was shown as the distance from tear drop, and cup medialization was shown as horizontal distance from inner wall of acetabulum. Leg elongation and offset change were navigated by comparing the two reference points in femur between registration before neck resection and that after inserting the trial implant. After operation, the cup angles were measured on CT image, and cup center position, leg elongation and offset change were measured on plain radiography. We compared these values that indicated by the navigation system to those measured on the CT image and the plain radiography. The average cup inclination was 37.5 ± 7.0 degree and anteversion was 22.2 ± 4.7 degree. The average absolute difference between navigation and measured angles were 5.2 ± 4.0 degree in inclination, 5.9 ± 4.0 degree in anteversion. The difference of cup height was 5.8 ± 3.9 mm, cup medialization was 3.8 ± 2.7 mm, leg elongation was 4.3±3.3mm, and offset was 5.4±4.1mm, respectively. By using this new version navigation system, we can plan the cup center position and navigate it within smaller error of vertical and horizontal direction than the previous system. Moreover, leg elongation and offset change can be satisfactory navigated during operation. However surgeon's skill and learning curve might have influence the accuracy. We have to continue to evaluate this system and make effort to further improvement.
