Pore segmentation in electron micrographs: A probabilistic approach by ensemble machine learning

Publikation: Beitrag in FachzeitschriftArtikelForschungPeer-Review

Autorschaft

Externe Organisationen

  • Bundesanstalt für Geowissenschaften und Rohstoffe (BGR)
Forschungs-netzwerk anzeigen

Details

OriginalspracheEnglisch
Aufsatznummer108047
FachzeitschriftApplied clay science
Jahrgang279
Frühes Online-Datum14 Nov. 2025
PublikationsstatusVeröffentlicht - Jan. 2026

Abstract

In this study, an automatic method for segmenting pores in scanning electron microscopy images was developed. An ensemble of machine learning classifiers was combined with a fully connected conditional random field to obtain a spatial pore probability field. This field was then thresholded to produce coherent binary pore masks, and a confidence per pore Cl was defined to quantify the reliability of the segmentation. The approach was demonstrated on a broad-ion-beam polished sample of the shaley facies of the Opalinus Clay. Accurate segmentation enabled the derivation of pore size distributions (PSD), pore morphologies, orientations, and spatial statistics. By using the median of Cl per size range, a data-driven lower truncation limit for PSD fitting was established. The resulting microstructural metrics supported the interpretation of rock properties such as permeability. These results highlighted the method’s relevance for materials such as Opalinus Clay, which is investigated as a potential candidate for a host rock for nuclear waste storage.

ASJC Scopus Sachgebiete

Zitieren

Pore segmentation in electron micrographs: A probabilistic approach by ensemble machine learning. / Brysch, Marco; Laurich, Ben; Sester, Monika.
in: Applied clay science, Jahrgang 279, 108047, 01.2026.

Publikation: Beitrag in FachzeitschriftArtikelForschungPeer-Review

Brysch M, Laurich B, Sester M. Pore segmentation in electron micrographs: A probabilistic approach by ensemble machine learning. Applied clay science. 2026 Jan;279:108047. Epub 2025 Nov 14. doi: 10.1016/j.clay.2025.108047
Download
@article{39f206e80927475cb2a30ecd3b7639b3,
title = "Pore segmentation in electron micrographs: A probabilistic approach by ensemble machine learning",
abstract = "In this study, an automatic method for segmenting pores in scanning electron microscopy images was developed. An ensemble of machine learning classifiers was combined with a fully connected conditional random field to obtain a spatial pore probability field. This field was then thresholded to produce coherent binary pore masks, and a confidence per pore Cl was defined to quantify the reliability of the segmentation. The approach was demonstrated on a broad-ion-beam polished sample of the shaley facies of the Opalinus Clay. Accurate segmentation enabled the derivation of pore size distributions (PSD), pore morphologies, orientations, and spatial statistics. By using the median of Cl per size range, a data-driven lower truncation limit for PSD fitting was established. The resulting microstructural metrics supported the interpretation of rock properties such as permeability. These results highlighted the method{\textquoteright}s relevance for materials such as Opalinus Clay, which is investigated as a potential candidate for a host rock for nuclear waste storage.",
keywords = "Machine learning, Opalinus Clay, Pore segmentation, Scanning Electron Microscopy, Uncertainty quantification",
author = "Marco Brysch and Ben Laurich and Monika Sester",
note = "Publisher Copyright: {\textcopyright} 2025 The Authors.",
year = "2026",
month = jan,
doi = "10.1016/j.clay.2025.108047",
language = "English",
volume = "279",
journal = "Applied clay science",
issn = "0169-1317",
publisher = "Elsevier BV",

}

Download

TY - JOUR

T1 - Pore segmentation in electron micrographs

T2 - A probabilistic approach by ensemble machine learning

AU - Brysch, Marco

AU - Laurich, Ben

AU - Sester, Monika

N1 - Publisher Copyright: © 2025 The Authors.

PY - 2026/1

Y1 - 2026/1

N2 - In this study, an automatic method for segmenting pores in scanning electron microscopy images was developed. An ensemble of machine learning classifiers was combined with a fully connected conditional random field to obtain a spatial pore probability field. This field was then thresholded to produce coherent binary pore masks, and a confidence per pore Cl was defined to quantify the reliability of the segmentation. The approach was demonstrated on a broad-ion-beam polished sample of the shaley facies of the Opalinus Clay. Accurate segmentation enabled the derivation of pore size distributions (PSD), pore morphologies, orientations, and spatial statistics. By using the median of Cl per size range, a data-driven lower truncation limit for PSD fitting was established. The resulting microstructural metrics supported the interpretation of rock properties such as permeability. These results highlighted the method’s relevance for materials such as Opalinus Clay, which is investigated as a potential candidate for a host rock for nuclear waste storage.

AB - In this study, an automatic method for segmenting pores in scanning electron microscopy images was developed. An ensemble of machine learning classifiers was combined with a fully connected conditional random field to obtain a spatial pore probability field. This field was then thresholded to produce coherent binary pore masks, and a confidence per pore Cl was defined to quantify the reliability of the segmentation. The approach was demonstrated on a broad-ion-beam polished sample of the shaley facies of the Opalinus Clay. Accurate segmentation enabled the derivation of pore size distributions (PSD), pore morphologies, orientations, and spatial statistics. By using the median of Cl per size range, a data-driven lower truncation limit for PSD fitting was established. The resulting microstructural metrics supported the interpretation of rock properties such as permeability. These results highlighted the method’s relevance for materials such as Opalinus Clay, which is investigated as a potential candidate for a host rock for nuclear waste storage.

KW - Machine learning

KW - Opalinus Clay

KW - Pore segmentation

KW - Scanning Electron Microscopy

KW - Uncertainty quantification

UR - http://www.scopus.com/inward/record.url?scp=105021857811&partnerID=8YFLogxK

U2 - 10.1016/j.clay.2025.108047

DO - 10.1016/j.clay.2025.108047

M3 - Article

AN - SCOPUS:105021857811

VL - 279

JO - Applied clay science

JF - Applied clay science

SN - 0169-1317

M1 - 108047

ER -

Von denselben Autoren