Related Experiment Video
Updated: Jul 9, 2026
07:38
Open-source Single-particle Analysis for Super-resolution Microscopy with VirusMapper
Published on: April 9, 2017
10.1K
Nano1D: An accurate computer vision software for analysis and segmentation of low-dimensional nanostructures
Ehsan Moradpur-Tari1, Sergei Vlassov2, Sven Oras2
1Institute of Technology, University of Tartu, Nooruse 1, 50411 Tartu, Estonia.
Ultramicroscopy
|March 19, 2024
Summary
A new physics-based model, Nano1D, enables accurate, automated quantitative analysis of 1D nanoparticles in microscopy images. It excels at segmenting and measuring overlapping nanoparticles, overcoming limitations of current methods.
Area of Science:
- Materials Science
- Image Analysis
- Computational Modeling
Background:
- Manual and qualitative analysis of nanoparticles in microscopy images is time-consuming and prone to error.
- Existing computational models struggle with accurate segmentation and analysis of overlapping 1D nanostructures.
- There is a significant need for automated, quantitative methods for nanoparticle characterization.
Purpose of the Study:
- To develop and validate a physics-based computational model, Nano1D, for autonomous quantitative analysis of 1D deformable, overlapping objects in microscopy images.
- To accurately segment and measure geometrical characteristics of nanoparticles, including length and diameter.
- To provide a robust solution for analyzing various 1D nanostructures and microstructural features.
Main Methods:
- The Nano1D model employs a four-step process: preprocessing, segmentation, separation of overlapped objects, and geometrical measurements.
- The model was tested on Scanning Electron Microscopy (SEM) images of silver (Ag) and gold (Au) nanowires, and fragmented Ag nanowires.
- Analysis included objects with varying sizes, densities, orientations, and degrees of overlap.
Main Results:
- Nano1D successfully segmented and analyzed geometrical characteristics (length, average diameter) of 1D nanoparticles with over 99% accuracy.
- The model demonstrated robustness, unaffected by object size, number, density, orientation, or overlap.
- Performance significantly surpassed current machine learning and computational models, particularly in handling overlapping objects.
Conclusions:
- The Nano1D model offers a highly accurate and automated solution for quantitative analysis of 1D nanoparticles and microstructural features.
- Its ability to precisely segment and analyze overlapping objects represents a significant advancement over existing methodologies.
- The user-friendly graphical interface makes Nano1D applicable to a wide range of 1D nanostructures and microstructural elements.
