Virtual facial reconstruction based on accurate registration and fusion of 3D facial and MSCT scans.

Abstract

Accurate registration and feasible fusion of a three-dimensional (3D) photorealistic surface images (captured using the FaceSCAN3D® Scientific Photo Lab) from multislice spiral computed tomography (MSCT) images is important to achieve optimal craniomaxillofacial surgery outcomes; thus, the aim of this work was to optimize this process. MSCT and 3D facial scans were acquired from 37 randomly selected patients. Using the Invesalius software package, skin and bone tissue models were reconstructed. The 3D photorealistic surface images were then constructed. Four image registration processes were performed using Geomagic Studio software: manual Procrustes and semi-automatic registration (global modified ICP) using both 7 and 15 anthropometric landmarks. The statistical differences were assessed using one-way ANOVA and LSD tests. P values <0.05 were considered significant. Average distances between these two surfaces measured 0.99 (SD 0.13 mm), 0.77 (SD 0.11 mm), 0.99 (SD 0.15 mm), and 0.77 mm (SD 0.10 mm) via the four image registration methods. Statistical differences were observed between the manual and semi-automatic registration groups (according to 7 and 15 anthropometric landmarks, respectively). Image registration errors for the entire virtual face were <0.8 mm in the semi-automatic registration groups. Employing the Procrustes registration system using seven anthropometric landmarks together with global registration allows accurate registration and feasible fusion of 3D facial scans with reconstructed 3D MSCT image data.