Publication Details

Graph-based deep learning segmentation of EDS spectral images for automated mineral phase analysis

JURÁNEK Roman, VÝRAVSKÝ Jakub, KOLÁŘ Martin, MOTL David and ZEMČÍK Pavel. Graph-based deep learning segmentation of EDS spectral images for automated mineral phase analysis. Computers and Geosciences, vol. 165, no. 8, 2022, pp. 1-2. ISSN 0098-3004. Available from: https://www.sciencedirect.com/science/article/pii/S0098300422000668
Czech title
Grafová segmentační metoda pro automatickou analýzu minerálních fází
Type
journal article
Language
english
Authors
Juránek Roman, Ing., Ph.D. (DCGM FIT BUT)
Výravský Jakub (TESCAN GROUP, a.s)
Kolář Martin, Ph.D. (DCGM FIT BUT)
Motl David, Ing. (FEECS BUT)
Zemčík Pavel, prof. Dr. Ing. (DCGM FIT BUT)
URL
Keywords

Segmentation, Deep learning, EDS spectra, Automated mineralogy

Abstract

We introduce a novel method for graph-based segmentation of spectral images obtained using a Scanning Electron Microscope (SEM) equipped with an Energy Dispersive X-ray spectroscopy (EDS) detector. The method exploits deep learning along with fusion of rasterized electron microscopy images with sparse EDS samples to obtain accurate mineralogy segmentation with high efficiency. Improvements over previous methods are with respect to the goal of an improved quantitative and qualitative assessment of segmentation, so that volumetric composition is indirectly addressed. We describe the principles of the novel method, show experimental results on real samples and demonstrate its advantages in comparison to the state of the art. The new method performs unsupervised clustering on sparsely measured EDS spectra, allowing for classification of unseen mineralogical compounds. Then, the processed spectra are combined with single channel SEM measurements through an optimized lattice, where a Markov Field is used to perform spatial segmentation in image. The benefit of this material-agnostic method is that clusters can then be (separately) classified, analyzed, and small grains with distinct EDS measurements are more accurately separated than in previous methods. These improved results are evaluated quantitatively on ground-truth electron microscope measurements with dense high-count EDS data, as well as visually through analysis by a mineralogist.

Published
2022
Pages
1-2
Journal
Computers and Geosciences, vol. 165, no. 8, ISSN 0098-3004
Publisher
Elsevier Science
DOI
UT WoS
000817165900005
EID Scopus
BibTeX
@ARTICLE{FITPUB12946,
   author = "Roman Jur\'{a}nek and Jakub V\'{y}ravsk\'{y} and Martin Kol\'{a}\v{r} and David Motl and Pavel Zem\v{c}\'{i}k",
   title = "Graph-based deep learning segmentation of EDS spectral images for automated mineral phase analysis",
   pages = "1--2",
   journal = "Computers and Geosciences",
   volume = 165,
   number = 8,
   year = 2022,
   ISSN = "0098-3004",
   doi = "10.1016/j.cageo.2022.105109",
   language = "english",
   url = "https://www.fit.vut.cz/research/publication/12946"
}
Back to top