5: Correlation Of Face Transplant Smile Excursion Measurements With Emotional Evaluation By Artificial Intelligence

Purpose: Examining the outcomes and progression of face transplants has expanded to include the recent advancements in artificial intelligence. This novel methodology has allowed for the objective exploration of emotional expression. To our knowledge, this examined emotional expression has yet to be correlated to the hypothesized functioning of the allograft. We have combined the result of the AI software with a software analyzing smile excursion; demonstrating the progression of the allograft in our face transplant patient. This study helps illuminate the software possibilities for objective measures in transplant progression. Methods: Images were analyzed from our face transplant patient post-operatively at approximately 6-month intervals for four years. The still images for each time interval include both a neutral facial expression and a full-face smile with teeth. The images were taken from video clips of the patient and then analyzed using the FACEgram software (Hadlock & Urban, 2012). The measurements of the smile excursion were standardized based off of the pupil diameter and excursion was measured relative to the neutral expression of each time period. The emotional expression data was acquired using FaceReader AI software from Noldus Information Technology (Wageningen, The Netherlands). The emotional analysis was done on 2 second clips from clinical videos of the patient, the same videos used for the still images. This data was correlated with the smile excursion measurements. Results: The post-operative smile excursion analysis from 6 to 49 months demonstrated a near double in smile excursion from 3.58mm to 6.04mm. The smile excursion data has a squared correlation of 0.705; a strong upward linear trend of excursion. From 6 to 35 months, the happy emotion increased gradually from 0% to 15.8% (49-month data unavailable). Performed correlative analysis showed a positive correlation coefficient of 0.63. This moderate to strong correlation demonstrates the intuitive relationship of happy emotion and lip excursion during full face smiling. The findings also substantiate the AI software’s detection of increasing activity in the lip corner puller action unit from zero to 3/5 intensity over the same period. Conclusion: This study demonstrates the correlation of smile excursion to happy emotional expression in our face transplant patient. The correlation square of the lip excursion shows a trend of increased functioning of the allograft over time; substantiating the AI’s finding. This data provides evidence for the use of artificial intelligence as a measure of transplant strength progression. The conjoined analysis of patient images and videos has supported proof for the ability to analyze transplant healing progression and efficacy in an objective form. Hadlock, T. A., & Urban, L. S. (2012). Toward a Universal, Automated Facial Measurement Tool in Facial Reanimation. Archives of Facial Plastic Surgery,14(4). doi:10.1001/archfacial.2012.111