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 Concept Videos

You might also read

Related Articles

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

Sort by
Same author

Granger-Causality-Based Multi-Frequency Band EEG Graph Feature Extraction and Fusion for Emotion Recognition.

Brain sciences·2022
Same author

Diagnostic Predictive Value of Tryptase, Serum Amyloid A and Lipoprotein-Associated Phospholipase A2 Biomarker Groups for Large Atherosclerotic Cerebral Infarction.

Emergency medicine international·2022
See all related articles

Related Experiment Video

Updated: Oct 20, 2025

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

3.0K

Dual-Path Attention Compensation U-Net for Stroke Lesion Segmentation.

Haisheng Hui1, Xueying Zhang1, Zelin Wu1

  • 1College of Information and Computer, Taiyuan University of Technology, Taiyuan 030024, China.

Computational Intelligence and Neuroscience
|September 10, 2021
PubMed
Summary

A new dual-path attention compensation U-Net (DPAC-UNet) improves stroke lesion segmentation accuracy by using an auxiliary network to correct attention errors. This method significantly enhances segmentation performance on MRI scans.

More Related Videos

Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application
05:56

Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application

Published on: April 14, 2023

2.8K
Dual-Task Stroop Paradigm for Detecting Cognitive Deficits in High-Functioning Stroke Patients
07:42

Dual-Task Stroop Paradigm for Detecting Cognitive Deficits in High-Functioning Stroke Patients

Published on: December 16, 2022

3.2K

Related Experiment Videos

Last Updated: Oct 20, 2025

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

3.0K
Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application
05:56

Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application

Published on: April 14, 2023

2.8K
Dual-Task Stroop Paradigm for Detecting Cognitive Deficits in High-Functioning Stroke Patients
07:42

Dual-Task Stroop Paradigm for Detecting Cognitive Deficits in High-Functioning Stroke Patients

Published on: December 16, 2022

3.2K

Area of Science:

  • Medical Imaging
  • Artificial Intelligence
  • Neurology

Background:

  • Attention U-Net models enhance medical image segmentation by focusing on salient features.
  • Challenges remain in segmenting small or blurred stroke lesions due to attention U-Net's potential for incorrect attention maps.

Purpose of the Study:

  • To develop an improved segmentation method for stroke lesions that addresses the limitations of standard attention U-Net.
  • To enhance the accuracy and robustness of stroke lesion segmentation, particularly for challenging cases.

Main Methods:

  • Proposed a dual-path attention compensation U-Net (DPAC-UNet) with identical primary and auxiliary attention U-Net structures.
  • The primary network performs segmentation, while the auxiliary network generates compensation coefficients to correct attention errors.
  • Introduced weighted binary cross-entropy Tversky (WBCE-Tversky) loss for primary network training and tolerance loss for auxiliary network training.

Main Results:

  • DPAC-UNet achieved a 6% higher Dice Similarity Coefficient (DSC) score compared to single-path attention U-Net.
  • The proposed method outperformed existing segmentation techniques on the ATLAS dataset.
  • Demonstrated superior performance in segmenting stroke lesions from MRI scans.

Conclusions:

  • DPAC-UNet effectively compensates for attention coefficient errors, leading to more accurate stroke lesion segmentation.
  • The dual-path architecture and novel loss functions provide a powerful tool for medical image analysis.
  • This method shows significant potential for clinical application in stroke diagnosis and treatment planning.