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 Videos

DeepEM3D: approaching human-level performance on 3D anisotropic EM image segmentation.

Tao Zeng1, Bian Wu1, Shuiwang Ji1

  • 1School of Electrical Engineering and Computer Science, Washington State University, Pullman, WA 99164, USA.

Bioinformatics (Oxford, England)
|April 6, 2017
PubMed
Summary
This summary is machine-generated.

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

Postoperative anlotinib plus radiotherapy in patients with newly diagnosed, unmethylated O<sup>6</sup>-methylguanine-DNA methyltransferase glioblastoma: A single-arm, phase 2 study.

Cancer·2026
Same author

TDLAS-based distributed monitoring method for multi-component gas leakage in crude oil storage tanks.

Optics express·2026
Same author

Global trends in robotic-assisted laparoscopic surgery (RALS): a bibliometric analysis (2000-2024).

Updates in surgery·2026
Same author

Activation of PPARγ and CPT1A Mediates the Hepatoprotective Effect of Ginsenoside CK against NAFLD in Rats.

Biological procedures online·2026
Same author

PythiaStudio: a one-stop protein engineering platform powered by Pythia model suite.

Nucleic acids research·2026
Same author

Should standard concurrent chemoradiotherapy remain the preferred treatment for elderly patients with locoregionally advanced nasopharyngeal carcinoma?

Radiation oncology (London, England)·2026

DeepEM3D is a deep learning method for segmenting 3D electron microscopy images, achieving near human-level performance in neurite reconstruction. This advancement is crucial for high-throughput neuroscience research.

Area of Science:

  • Neuroscience
  • Computational Biology
  • Image Analysis

Background:

  • High-throughput 3D electron microscopy (EM) demands automated image analysis for neuroscience.
  • Current automated EM image segmentation for neurite reconstruction lags behind human performance.

Purpose of the Study:

  • To develop an automated deep learning method for segmenting 3D anisotropic brain EM images.
  • To improve the efficiency and reliability of neurite reconstruction in EM data.

Main Methods:

  • Proposed DeepEM3D, a deep learning model for 3D anisotropic EM image segmentation.
  • Incorporated multi-scale contextual information and novel boundary map generation.
  • Utilized optimized model ensembles to address segmentation challenges.

Related Experiment Videos

Main Results:

  • DeepEM3D achieved the #1 rank in the SNEMI3D challenge (Oct 2016).
  • Performance was nearly indistinguishable from human-level accuracy (Rand error 0.06015 vs. 0.05998).

Conclusions:

  • DeepEM3D offers a highly efficient and reliable solution for EM image segmentation.
  • The method significantly advances automated neurite reconstruction in neuroscience research.