Guided surgery using rapid prototyping patient-specific guides: A methodology to quantify mechanical stability and uniqueness of fit

Rapid prototyping technology (3-dimensional printing) in conjunction with the virtual manipulation of computer-generated 3-D anatomical models rendered from diagnostic images (CT and MRI) is a novel technique for directing image-guided surgery. By incorporating a graphical user interface, the surgeon can plan an intracranial surgical procedure by fabricating a patient-specific reference frame (mask-like facial frame) through the use of rapid prototyping technology.; Early testing of these patient-specific frames revealed the necessity to include a step prior to fabrication to analyze the design of the frame for stability of fit. In response to this need, the objective in the present investigation is to develop a method that numerically assesses the frame's stability. The approach was to create a finite-element simulation that models the patient-specific reference frame as a rigid layer composed of linear tetrahedron elements under a deformable material simulated by a bed of linear spring elements. The program analyzes the displacement of the rigid layer by simulating external forces distributed around its perimeter. The stability of fit is characterized by the response of the linear springs.; Three patients with different facial anatomical features were selected as training data for testing the simulation program. For each case two patient-specific reference frames, one that involves more surface area and surface variations such as contours of the nose bridge and another that consists of less surface area and less surface variations, were built with Rapid prototyping based on the patient's 3-D anatomical facial models. In all three cases the methodology correlated a stability score with the stability of the real reference frames. In addition two more cases were evaluated; the first did not incorporate the facial structures and was design to hold onto the top of the head in a cap-like manner, the second case was a virtually created half sphere. In both cases it was expected to find no a unique fit, and correspondingly the simulation found both cases to be unstable.