Towards automatic computer-aided planning in arthroplasty surgery by innovative methods for processing the bone surface models

Abstract

2D and 3D-based innovative methods for surgical planning and simulation systems in orthopedic surgery have emerged enabling the interactive or semiautomatic identification of the clinical landmarks (CL) on the patient individual virtual bone anatomy. They allow the determination of the optimal implant sizes and positioning according to the computed CL, the visualization of the virtual bone resections, and the simulation of the overall intervention prior to surgery. Such innovative methods allow designing personalized resection guides, which substitute the traditional jigs and avoid any other alignment instrument and even navigation support. The virtual palpation of CL, highly dependent upon examiner’s expertise, was proved to be time consuming and suffered from considerable inter-observer variability. In this contribute, we disclose a fully automatic algorithmic framework that processes the femur surface, integrating surface curvature analysis, quadric fitting, recursive clustering, and clinical knowledge, aiming at computing the main parameters femur CL, namely the femoral shaft (FDA), transepicondylar (TEA), anterior–posterior (WL), posterior condylar (PCL) axes. At highest surface resolutions, the relative median error in the direction of the FDA, AFA, PCL, WL, and TEA was less than 0.50°, 1.20°, 1.0°, 1.30°, 1.50°, respectively. As expected, at the lowest surface resolution, the repeatability decreased to 1.20°, 2.70°, 3.30°, 3.0°, 4.70°, respectively. The computed directions of the FDA, PCL, WL, and TA were in agreement (0.60°, 1.55°, 1.90°, 2.40°) with the corresponding reference parameters manually identified in the original CT images by medical experts and with literature. We summarize that: (a) the AFA can be robustly computed by a geometrical analysis of the posterior profiles of the two condyles, and it can be considered a useful alternative to the TEA; (b) higher surface resolutions lead to higher repeatability of all computed quantities; (c) the TEA is less repeatable of the other axes. In conclusion, the method does not require any manual initialization, it can be automatically applied to the left and right surfaces, it is independent of the coordinate system of the original CT datasets, it is independent of the scale of the surface, and the algorithms show high stability and reduced computational load.