Related Experiment Video
Updated: May 16, 2026
11:30
Freeze-Fracture Electron Microscopy for Extracellular Vesicle Analysis
Published on: September 16, 2022
VesiclePy: A machine learning vesicle analysis toolbox for volume electron microscopy
Jason Ken Adhinarta1, Yutian Fan1, Adam Gohain1
1Computer Science Department, Boston College, Chestnut Hill, Massachusetts, United States of America.
Plos Computational Biology
|May 14, 2026
Summary
Researchers developed VesiclePy, a new software pipeline for automatically analyzing neuronal vesicles in large electron microscopy datasets. This tool aids in understanding neuronal communication by precisely mapping and classifying these critical cellular components.
Area of Science:
- Neuroscience
- Cell Biology
- Computational Biology
Background:
- Vesicles are essential for neuronal communication, packaging neurotransmitters and neuropeptides.
- Analyzing the full complement of vesicles within neuronal morphology is challenging due to their small size.
- Volume electron microscopy (vEM) offers nanoscale resolution but generates large datasets, complicating vesicle analysis.
Purpose of the Study:
- To develop an integrated computational pipeline, VesiclePy, for automated segmentation, classification, and spatial analysis of vesicles in large-volume vEM data.
- To address the challenges of processing and analyzing tens of thousands of vesicles in 3D.
- To enable quantitative spatial analysis of vesicles relative to neuronal structures.
Main Methods:
- Development of VesiclePy, a software package integrating deep learning and human proofreading for vesicle analysis.
- Implementation of chunked processing for efficient handling of large vEM datasets.
- Utilizing high-pressure frozen serial EM data of Hydra vulgaris for demonstration and validation.
Main Results:
- VesiclePy successfully processed a multi-terabyte serial EM dataset, annotating 53,851 vesicles from 20 neurons.
- The pipeline classified vesicles into 5 distinct types, providing unique IDs and 3D locations.
- Quantitative clustering of neurons into subtypes was achieved by combining vesicle and morphological data.
Conclusions:
- VesiclePy provides a streamlined and efficient solution for automated vesicle analysis in large-volume vEM data.
- The tool facilitates detailed spatial analysis of vesicles and their relationship to neuronal targets.
- VesiclePy enables quantitative subtyping of neurons based on their vesicle content and morphology.
Related Concept Videos
01:38Intralumenal Vesicles and Multivesicular Bodies
Intraluminal vesicles (ILVs) are small vesicles 50-80 nm in diameter formed during the maturation of early endosomes. A specialized endosome containing numerous ILVs is called a multivesicular body (MVB). ILVs contain internalized molecules such as antigens, nucleic acids, proteins, and metabolites. Some of these molecules are released from the MVBs inside exosomes and are transported to other cells. Other MVBs contain molecules that are retained in the ILVs and are later degraded within the...
01:32Pinching-off of Coated Vesicles
Vesicle budding is orchestrated by distinct cytosolic proteins such as adaptor proteins, coat proteins, and GTPases. To initiate vesicle budding, membrane-bending proteins containing crescent-shaped BAR domains bind to the lipid heads in the bilayer and distort the membrane to form a protein-coated vesicle bud. Adaptors proteins such as AP2 for clathrin-coated vesicles can nucleate on the deformed membrane. Finally, coat proteins such as clathrin or COPI and COPII assemble into a coat forming...