Digitizing Eyelid Contour: Image-Derived Geometric Biomarkers for Objective Ptosis Severity Assessment

Abstract Purpose To introduce and evaluate three novel two-dimensional (2D) geometric eyelid contour indices—Upper Nasal Index (UNI), Upper Temporal Index (UTI), and Upper Lid Contour Circularity Index (ULCCI)—for objective blepharoptosis severity assessment, bridging expert clinical impression with traditional margin reflex distance 1 (MRD1)-based classification. Methods This cross-sectional study evaluated 94 eyes from 94 participants (eyes stratified into control/no ptosis, mild, moderate, and severe groups) using both expert clinical grading (holistic visual impression) and standard MRD1 measurements. Standardized full-face digital photographs were processed using MATLAB to calculate 2D area-derived regional indices (UNI and UTI) and ULCCI, defined as the ratio of upper eyelid contour length to the circumference of the upper half of a reference circle. Group differences were assessed using generalized linear models, including analyses adjusted for age and sex, and diagnostic performance was evaluated using ROC curve analysis. Results Expert clinical grading demonstrated substantial agreement with MRD1 classification (weighted Cohen's κ = 0.775, 95% CI 0.695–0.855) and showed excellent reliability (interobserver ICC = 0.92; intraobserver ICC = 0.93). All three indices significantly differentiated overall severity groups under both classification frameworks after adjustment for age and sex (all P < 0.001). UNI demonstrated the greatest numerical AUC for differentiating no ptosis from mild ptosis in both expert clinical grading (AUC = 0.768) and MRD1-based classification (AUC = 0.846). UTI demonstrated the greatest numerical AUC for differentiating moderate from severe ptosis using MRD1-based classification (AUC = 0.817), while ULCCI showed the greatest numerical AUC under expert clinical grading for this comparison (AUC = 0.744). Conclusions Novel 2D geometric indices provide objective quantitative measures of eyelid contour that complement conventional central MRD1 assessment. UNI showed particular utility in distinguishing early ptotic changes, whereas UTI quantified advanced temporal contour changes and ULCCI reflected global contour characteristics that contribute to expert clinical assessment. Together, these digital biomarkers provide a reproducible framework for quantitative eyelid phenotyping and may support future scalable image-based diagnostic documentation, automated assessment workflows, and computational approaches to personalized ophthalmic care.