Abstract
Accurate and automated characterization of surface wettability is essential for optical sensing, material inspection, and polymer surface engineering. Conventional contact angle (CA) measurements rely on 2D side-view imaging and manual contour fitting, which introduce user-dependent variability and limit throughput. This study presents an intelligent optical coherence tomography (OCT)-based sensing framework that integrates OCT with a convolutional neural network (CNN) regression for simultaneous estimation of four CA metrics—left, right, circular, and elliptical—directly from volumetric cross-sectional data. Multi-output and single-output regression models were trained and compared in terms of accuracy and computational efficiency. The multi-output model achieved R 2 > 0.92 across all CA types, while single-output models achieved R 2 = 0.93–0.96 with reduced variance. Experimental validation confirmed that the proposed OCT-enabled CNN framework accurately resolves the droplet–surface interface and meniscus curvature without manual edge detection, geometric fitting, or engineered features. The results demonstrate a fully automated, depth-resolved, and high-throughput optical sensing approach for quantitative wettability characterization of polymeric materials, offering a robust solution for next-generation intelligent surface measurement systems.
Funding Information
  • Fulbright Association (FY-2023-TR-PD-01)

This publication has 37 references indexed in Scilit: