Related Experiment Video
Updated: Aug 19, 2025
10:25
Deep Learning-Based Segmentation of Cryo-Electron Tomograms
Published on: November 11, 2022
9.1K
Deep neural network automated segmentation of cellular structures in volume electron microscopy
Benjamin Gallusser1,2, Giorgio Maltese1, Giuseppe Di Caprio1,3
1Program in Cellular and Molecular Medicine, Boston Children's Hospital, Boston, MA.
The Journal of Cell Biology
|December 5, 2022
Summary
Automated segmentation of intracellular substructures in electron microscopy (ASEM) uses AI to identify cellular components. This method speeds up analysis of electron microscopy data, enabling new biological discoveries.
Area of Science:
- Cell Biology
- Biophysics
- Computational Biology
Background:
- Volume electron microscopy is crucial for cell biology.
- Manual identification of intracellular structures is time-consuming and limits data analysis.
Purpose of the Study:
- To develop an automated pipeline for segmenting intracellular substructures in electron microscopy data.
- To overcome the bottleneck of manual structure identification.
Main Methods:
- Developed Automated Segmentation of Intracellular Substructures in Electron Microscopy (ASEM) pipeline.
- Utilized convolutional neural networks trained on limited annotated images.
- Employed a model refinement strategy with additional annotations for improved generalization.
Main Results:
- Successfully identified diverse intracellular structures including mitochondria, Golgi apparatus, endoplasmic reticulum, nuclear pore complexes, caveolae, clathrin-coated pits, and vesicles.
- Revealed variability in membrane-nuclear pore diameters within a single cell.
- Derived morphological metrics for clathrin-coated pits and vesicles.
Conclusions:
- ASEM significantly accelerates the analysis of electron microscopy data.
- The developed method allows for detailed morphological analysis of cellular structures.
- Findings support established models of clathrin-coated pit and vesicle formation.
