Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Video

Updated: May 28, 2026

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
04:48

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

A study on a high efficiency phase organization segmentation model for SEM images based on deep learning.

Chao Wang1,2, Changhao Wang1, Zhipeng Chang1

  • 1Key Laboratory of Advanced Functional Materials of Education Ministry of China, College of Materials Science and Engineering, Beijing University of Technology, Beijing 100124, China. wangch33@bjut.edu.cn.

Nanoscale
|May 26, 2026
PubMed
Summary

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

The study of hydrogen adsorption-induced topological surface state in-out hop in MgB<sub>2</sub> nodal-line semimetals <i>via</i> physics-informed Bayesian optimization.

Physical chemistry chemical physics : PCCP·2026
Same author

Tailoring Superlattice Dimensions: A Pathway to Emergent Quantum Functional Devices.

Small (Weinheim an der Bergstrasse, Germany)·2025
Same author

A First-Principles Calculation Study of the Catalytic Properties of Two-Dimensional Bismuthene Materials for Carbon Dioxide Reduction.

Materials (Basel, Switzerland)·2025
Same author

Retina-Inspired Flexible Visual Synaptic Device for Dynamic Image Processing.

ACS applied materials & interfaces·2025
Same author

Variation of topological surface states of nodal line semimetal MgB<sub>2</sub> resulting from adsorption of hydrogen, hydroxide, and water molecules.

Physical chemistry chemical physics : PCCP·2024
Same author

Synergistic Assistance of Ir Clusters and NiCo<sub>2</sub>O<sub>4</sub> Nanosheets Interfaces in Direct O-O Coupling for High-Efficiency Alkaline Oxygen Evolution.

ACS applied materials & interfaces·2024
This summary is machine-generated.

A new deep learning model, DNU-Net, precisely segments microstructures in scanning electron microscopy (SEM) images, even with limited data. This robust AI approach enhances material characterization for advanced alloy design.

Area of Science:

  • Materials Science
  • Artificial Intelligence
  • Image Analysis

Background:

  • Quantitative analysis of material microstructures is vital for understanding material properties.
  • Traditional U-Net models struggle with noise and overfitting in scanning electron microscopy (SEM) image segmentation, especially with small datasets.
  • Accurate segmentation of microstructural phases is essential for materials characterization.

Purpose of the Study:

  • To develop a novel deep learning model, DNU-Net, for robust and accurate SEM image segmentation of material microstructures.
  • To improve the efficiency, robustness, and generalizability of microstructural phase segmentation, particularly for limited training data.
  • To enable precise quantitative analysis of complex microstructures in materials like Ti-6Al-4V (TC4) titanium alloy.

Main Methods:

Related Experiment Videos

Last Updated: May 28, 2026

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
04:48

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

  • Proposed a novel deep learning model, DNU-Net, integrating a denoising layer and dropout regularization into the U-Net architecture.
  • Applied DNU-Net to segment microstructural phases in scanning electron microscopy (SEM) images of Ti-6Al-4V (TC4) titanium alloy.
  • Evaluated segmentation performance using metrics such as mean intersection over union (IoU) and pixel accuracy (PA), and assessed robustness under noise and varying illumination.

Main Results:

  • DNU-Net achieved high segmentation accuracy for Ti-6Al-4V (TC4) titanium alloy, with mean IoU of 90.56% and PA of 98.46%.
  • The model demonstrated robustness against noise and challenging illumination conditions, with quantitative analysis error below 5%.
  • DNU-Net outperformed conventional digital image processing methods in efficiency, precision, and noise sensitivity.

Conclusions:

  • The DNU-Net framework provides an efficient, robust, and generalizable solution for microstructural phase segmentation in SEM images, even with limited data.
  • The model enables precise quantitative analysis of complex microstructures, offering a significant advancement in material characterization.
  • The proposed methodology is transferable across various alloy and material systems, supporting AI-driven materials design.