Biometric facial recognition after bimaxillary orthognathic surgery

Orthognathic surgery is performed to correct dentoskeletal deformities and restore maxillomandibular symmetry. These changes may affect biometric identification and verification. The aim of this study was to assess the facial recognition capability of a commercially available software application in patients who had undergone orthognathic surgery. This retrospective study included 24 patients, of both sexes, who were treated for dentofacial deformities: 12 skeletal Class II and 12 Class III. Images of these patients taken in neutral expression and in smiling expression, acquired preoperatively and at 6 months postoperative, were analysed using the software via a machine-learning tool through two neural network models. The software identified similarities between the preoperative and postoperative images based on Euclidian distances (ED). Regarding recognition in neutral facial expression, there was no significant difference in the mean ED between classes II and III (P = 0.50). However, in smiling facial expression, a significant difference was observed, with a higher mean ED in Class III (P = 0.009). Moreover, a significant difference between neutral and smiling facial expressions was observed for Class III patients (P = 0.023), but not for Class II patients (P = 0.65). Although orthognathic surgery can result in considerable changes to the face, the program under evaluation demonstrated excellent performance in recognizing patient facial similarities.