Orthognathic Surgery (OGS) is a routine treatment to correct dentofacial deformity which results in changes of patient’s facial form in all dimensions.
As Artificial intelligence (AI) prevails in the field of biometric facial recognition, knowledge on the ability of AI-based facial recognition to recognize patients who have undergone OGS should be kept in tandem with its rising reliance.
The objective of the study is to investigate if modern facial recognition software can detect facial changes in patients after OGS.
Pre- and post-surgery frontal photos were analyzed with four commercially available facial recognition software to produce facial Similarity Scores (SS) between the two sets of frontal photos.
Correlation with patient demographics and skeletal diagnosis, surgical type and magnitude, presence of asymmetry, and soft and hard tissue facial changes were analyzed.
Malocclusion is a prevalent condition worldwide. When associated with an underlying Skeletal Class 2 or 3 discrepancy, malocclusion may require treatment beyond orthodontic correction. Orthognathic Surgery (OGS) is a routine treatment to correct dentofacial deformity which results in changes of patient’s facial form in all dimensions. As Artificial intelligence (AI) prevails in the field of biometric facial recognition, knowledge on the ability of AI-based facial recognition to recognize patients who have undergone OGS should be kept in tandem with its rising reliance. The objective of the study is to investigate if modern facial recognition software can detect facial changes in patients after OGS. This cohort study included 39 patients who had undergone OGS and 39 control subjects, recruited from a tertiary center. Changes in hard and soft tissue landmarks of OGS patients were recorded using 3-dimensional (3D) stereophotogrammetry and lateral cephalograms. Pre- and post-surgery frontal photos were analyzed with four commercially available facial recognition software to produce facial Similarity Scores (SS) between the two sets of frontal photos. Differences between SS of study and control subjects were investigated. Correlation with patient demographics and skeletal diagnosis, surgical type and magnitude, presence of asymmetry, and soft and hard tissue facial changes were analyzed. There was a statistically significant difference in the SS of patients who underwent OGS compared to the controls across all four software (P < 0.001). Maxillary advancement surgery, changes to External Canthus-Alar (ExC-Alar) point, Paranasal point, A-point to Vertical Reference Plane (A-VRP) and hard tissue Pogonion point were found to significantly lower SS. Dlib face_recognition was most sensitive to post-surgery facial changes among the four software. Despite significant changes to the SS, the patients were successfully recognized as the same person by the software. However, patients with substantial increases in projection at the cheek, paranasal, and chin, are potentially vulnerable to recognition failure.
