Machine learning applications in SEM-based pore analysis: a review
Efi-Maria Papia, Alex Kondi, Vassilios Constantoudis
Abstract
Scanning Electron Microscopy (SEM) is a cornerstone technique for analyzing porous materials, providing high-resolution images essential for understanding material properties and performance. However, traditional SEM image analysis methods often involve manual interpretation and are limited by challenges such as noise, segmentation difficulties, and resolution constraints. Recent advancements in machine learning have revolutionized SEM image analysis, offering automated, accurate, and scalable solutions. These technologies enable precise pore size distribution measurement, pore shape classification, and network connectivity analysis while enhancing image quality through advanced denoising techniques. This paper reviews the integration of Artificial Intelligence (AI) in SEM-based porous material analysis, discussing its applications, challenges, and future directions. Through highlighting key contributions in the field, we aim to provide a comprehensive overview of how AI is reshaping SEM image analysis and unlocking new possibilities for porous material characterization, also emphasizing challenges and limitations that arise.