Effect of display modality on spatial accuracy of orthopaedic surgery pre-operative planning applications

Graphical representation of a patient's anatomy is fairly similar in the majority of orthopaedic surgery planning programs. The position of implantable devices is usually established using a three-pane window showing 2D cross sections of the CT data set taken on three user-selectable orthogonal planes. In some cases this orthogonal-plane representation is replaced or extended by interactive 3D visualization, obtained using surface rendering techniques. These ways to represent the CT data come naturally and easily to the programmer. However, the efficacy of these display strategies is questionable. The present study aims to assess if the display modality used to visualize CT data affects the inherent spatial accuracy achievable in a surgical planning application. A group of users was asked to perform repeatedly a specific planning task using various display interfaces to the same underlying software application. The planning task was designed to allow an assessment of the accuracy with which each user was able to position the prosthetic component. A specialized interface, called multimodal display, presented a peak error of 0.45 mm and 0.95 deg, significantly lower than the 2.4 mm and 4 deg for the othogonal slices interface, and the 3.8 mm and 16.7 deg for the 3D-rendering interface. The results of this study indicate an important effect of the type of visualization modality used to represent CT data on final accuracy of the planning operation.