Abstract
Introduction
Studies show that cup malpositioning using conventional techniques occurs in 50 to 74% of cases defined. Assessment of the utility of improved methods of placing acetabular components depends upon the accuracy of the method of measuring component positioning postoperatively. The current study reports on our preliminary experience assessing the accuracy of EOS images and application specific software to assess cup orientation as compared to CT.
Methods
Eighteen patients with eighteen unilateral THA had pre-operative EOS images were obtained for preoperative assessment of leg-length difference and standing pelvic tilt. All of these patients also had preoperative CT imaging for surgical navigation of cup placement. This allows us to compare cup orientation as measured by CT to cup orientation as measured using the EOS images.
Application specific software modules were developed to measure cup orientation using both CT and EOS images (HipSextant Research Application 1.0.13 Surgical Planning Associates Inc., Boston, Massachusetts). Using CT, cup orientation was determined by identifying Anterior Pelvic Plane coordinate system landmarks on a 3D surface model. A multiplanar reconstruction module allows for creation of a plane parallel with the opening plane of the acetabulum and subsequent calculation of plane orientation in the AP Plane coordinate space according to Murray's definitions of operative anteversion and operative inclination.
Using EOS DICOM images, spatial information from the images were used to reconstruct the fan beam projection model. Each image pair is positioned inside this projection model. Anterior Pelvic Plane coordinate points are digitized on each image and back-projected to the fan beam source. Corresponding beams are then used to compute the 3D intersection points defining the 3D position and orientation of the Anterior Pelvic Plane. Ellipses with adjustable radii were then used to define the cup border in each EOS image. By respecting the fan beam projection model, 3D planes defining the projected normal of the ellipse in each image are computed. 3D implant normal was estimated by determining 3D plane intersection lines for each image pair.
Implant center points are defined by using the back-projected and intersected ellipse center beams in the image pairs (Figure 1).
Results
The results are shown in Figure 2. The mean anteversion error was −0.9 degrees (SD 4.1, range −6.9 to 10.3). The mean inclination error was 1.8 (SD 2.1, range −2.9 to 8.6). All three cups with errors greater than 7 degrees were in cups with 40 or more degrees of anteversion.
Discussion and Conclusion
The current study, while very preliminary, demonstrates the potential that EOS images can be used to measure cup orientation with a reasonable degree of accuracy. Accurate determination of cup orientation appears to be more challenging in cups with higher anteversion.
