Unsupervised segmentation of irradiation$\unicode{x2010}$induced order$\unicode{x2010}$disorder phase transitions in electron microscopy

TL;DR

This paper introduces an unsupervised method for segmenting electron microscopy images using graph-based community detection, enabling real-time analysis of irradiation-induced phase transitions in materials.

Contribution

The authors develop a novel unsupervised segmentation approach combining CNN embeddings and community detection, suitable for real-time electron microscopy analysis.

Findings

Effective segmentation of irradiation-induced phase transitions.

Application to tracking amorphous fronts in thin films.

Potential for automated, real-time microscopy analysis.

Abstract

We present a method for the unsupervised segmentation of electron microscopy images, which are powerful descriptors of materials and chemical systems. Images are oversegmented into overlapping chips, and similarity graphs are generated from embeddings extracted from a domain$\unicode{x2010}$pretrained convolutional neural network (CNN). The Louvain method for community detection is then applied to perform segmentation. The graph representation provides an intuitive way of presenting the relationship between chips and communities. We demonstrate our method to track irradiation$\unicode{x2010}$induced amorphous fronts in thin films used for catalysis and electronics. This method has potential for "on$\unicode{x2010}$the$\unicode{x2010}$fly" segmentation to guide emerging automated electron microscopes.