Artificial intelligence–assisted quantitative evaluation of upper eyelid blepharoplasty outcomes in dermatochalasis: methodological validation and clinical correlation

BACKGROUND AND AIMS: This prospective study aimed to assess the functional and aesthetic outcomes of upper eyelid blepharoplasty in patients with dermatochalasis using a validated artificial intelligence (AI)–assisted morphometric analysis system and to determine its accuracy, reproducibility, and clinical applicability. METHODS: Twenty‑five patients (50 eyelids) with age‑related dermatochalasis underwent standardized upper eyelid blepharoplasty between April and November 2024. High‑resolution periocular photographs were obtained preoperatively and at 2 weeks postoperatively. Morphometric parameters—including MRD‑1, PFH, TPS, and BFS—were quantified automatically using a hybrid MediaPipe–OpenCV pipeline. Bilateral symmetry was quantified using an Asymmetry Index. Changes between pre- and postoperative asymmetry indices were analyzed using the Wilcoxon signed-rank test. Synthetic images were used solely for calibration, while all outcome measurements were derived from real patient photos. Accuracy was validated through manual measurements by two ophthalmologists, yielding an intraclass correlation coefficient (ICC) of 0.93 (p < 0.001). RESULTS: All patients completed follow‑up. The AI system demonstrated 97.5% accuracy and significant postoperative improvements across all parameters (p < 0.001) such as 86.7% precision, and a 92.9% F1-score in landmark detection, with a mean IoU of 0.82 and mAP > 0.80. Measurement variability decreased 68% versus manual methods. Area under the ROC curve (AUC) was 0.85 (95% CI 0.78–0.91), confirming strong discriminative power. CONCLUSION: AI‑assisted morphometric analysis allows objective, reproducible, and quantitative evaluation of blepharoplasty outcomes, minimizing observer bias and establishing standardized benchmarks for future surgical audits.