AI-Driven Image Analysis for Nanofiber Characterization: From Diameter Measurement to Multiparameter Assessment

The high surface area and porosity of the nanofibers enable a wide range of pharmaceutical applications, including tissue scaffolds, wound dressings, and drug-loaded films, as well as applications in other fields such as energy, electronics, and environmental remediation. The properties of nanofibers are directly dependent on the production parameters, such as applied voltage, solution flow rate, polymer concentration, solvent type, and collector distance, and there is a complex interplay between these parameters that makes their optimization challenging. Therefore, accurate determination of nanofiber properties, especially fiber diameter, is essential for quality control, process optimization, and functional performance assessment. This review systematically investigates computational methodologies employed in the characterization of nanofibers, with a particular focus on the measurement of fiber diameter. Initially, manual measurements and open-source tools such as DiameterJ, GIFT, and SIMpoly are described, highlighting their advantages and limitations. Subsequently, artificial intelligence-based strategies are described, ranging from classical machine learning models to deep learning architectures, as well as more advanced approaches such as generative frameworks and transformer-based models. In addition, comparisons with traditional characterization methods, industry applications including smart manufacturing, and automated quality control are outlined. Finally, the review examines emerging and prospective artificial intelligence methodologies in the analysis of nanofibers, offering conclusions and recommendations.