Computer aided design of cranial implants using deformable templates.

摘要

The design of patient-specific implants is a difficult task with traditional computer aided design (CAD) tools. I present a CAD approach that utilizes three dimensional (3D) computerized tomographic (CT) images for the design of cranial implants that are manufactured via a rapid prototyping (RP) device. The design process employing dedicated algorithms starts with a verbose patient's 3D cranial polygonal mesh image derived from 3D CT data followed by four design stages: First, the cranial defect margin is identified as a crestline on the patient's polygonal mesh image. Detection is assisted by a minimum-cost path searching-algorithm adapted from graph theory. The path traversal cost is primarily based on a principal curvature analysis. Second, either a left-right mirrored or an average skull template is interpolated into the defect site. The skull template deformation is driven by a two-pass thin plate spline (TPS) warp, first globally, then locally. Third, the skull template surface is regularly re-sampled and then smoothed using ray-surface collision detection and 2D low-pass image mask filtering. In addition, the portion of the skull template inside the cranial defect margin is partitioned from the rest of the skull template. Fourth, this prototype implant surface is validated by a check for non-intersection with adjacent patient skull and soft-tissue structures. The prototype implant surface is modified if necessary. An internal implant surface and a taper surface connecting the internal and external implant surfaces are interactively designed so as to insure appropriate implant-patient contact. This step results in a closed polygonal mesh surface that can be produced as a solid implant via an RP device. I tested this CAD approach with five patient cranial data sets with defects that had been previously repaired. The defects vary in size and location. The design environment was found to be reasonably interactive, requiring less than three hours following preparation of the patient's polygonal mesh skull image. These results suggest that the design procedure and cranial prosthetic implants resulting from my CAD approach would be clinically acceptable.