As further improvements in surgical accuracy are made possible by computer-aided surgery, there is a demand for new pre- and post- surgical assessment and more accurate intra-operative registration techniques. Ultrasonic palpation is being used in navigated hip surgery but as yet little work had been published on the identification of anatomical landmarks used in knee surgery with this technique. The aim of this study was to investigate the accuracy of the identification of the femoral condyles with ultrasound in both saline and in tissue mimicking material (TMM).

The system comprised of an image free navigation system (OrthoPilot, B Braun Aesculap) synchronized with a standard B-mode ultrasound system (Echoblaster 128, TELEMED) used with passive trackers. Bony anatomy was represented by two sawbone phantoms; one involving an isolated femur and one simulated knee joint. Both phantoms had fiducial markers in the form of steel pins inserted into the condylar eminences of the femur, providing sharply defined structural interfaces for determination of inter-condylar distance (ICD). Initial testing was completed in a waterbath filled with saline (NaCl 4500ppm) maintained at 22°C. Further testing used both sawbone phantoms encased in TMM. To gain accurate dimensions of the ICD, 3D models of both sawbone phantoms were created using a high-resolution non-contact 3D digitiser (Konica Minolta Sensing Inc.) and measurements taken using Geomagic software. Measurements for all test set-ups were repeated and mean (SD) values calculated.

The mean ICD measurement (SD) of the isolated femur from the high resolution 3D model was 53.6mm (1.2mm) (n=4). The ICD for the isolated femur in the saline water bath was 48.8mm (0.7mm) (n=5). For the isolated femur encased in TMM the mean ICD was 54.6mm (0.7mm) (n=4) with the probe positioned parallel to the shaft of the femur and 52.2mm (0.4mm) (n=5) with the probe held perpendicular to the femur. For the second phantom, which consisted of an articulated knee joint, the mean ICD measured from the high-resolution 3D model was 43.5mm (1.0mm) (n=5). When encased in TMM, the mean ICD derived from the navigation system was 42.6mm (1.4mm) (n=5).

Average ICD measurements for phantoms encased in TMM were within 1mm of that determined by high resolution, non-contact 3D digitization. However, results in the saline waterbath were less accurate, with an average difference of 4.8mm in ICD measurement. We believe these differences largely reflect the digitisation error associated with manual registration of the fiducial markers and highlights the difficulty in using this method and taking measurements within one scanned plane. Hence we are now developing a new method of automatic registration that uses multiple scans and will hopefully provide a more accurate outcome.