Leveraging Artificial Intelligence to Assess Perceived Age and Donor Facial Resemblance After Face Transplantation.

Abstract

Purpose: A major concern for patients undergoing facial transplantation relates to postoperative appearance. This study leverages artificial intelligence (AI) visual analysis software to provide an objective assessment of perceived age and degree of resemblance to the donor.

Methods: Postoperative images of 15 face transplant patients were analyzed by Visage Technologies Visage|SDK™ AI facial analysis software to determine perceived age. A subgroup of eight face transplant patients, for which donor and patient pretrauma photographs were available, was analyzed using the same software to determine the percent similarity match to the patients' postoperative image. Mann-Whitney and Wilcoxon rank sum tests were utilized to evaluate for perceived age and facial recognition matching percentage, respectively.

Results: AI perceived age was significantly more similar to the patient age (±3.5 years) than the donor age (±9.5, P = 0.0188). For facial resemblance, patients had a significantly higher average percent similarity match to their donor's face compared to their pretrauma native face (63% vs 57%, P = 0.0391).

Conclusions: Although patients more closely resembled their donor's resemblance posttransplantation, their perceived age correlated more significantly with their actual age than their donor allograft age. The findings of this study provide a helpful framework for counseling prospective patients on their expected appearance postoperatively.