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

Multi-modal Features Analysis and Performance Assessment for Endovascular Surgery Skills.

Annals of biomedical engineering·2026
Same author

Effect comparison between a new hypofractionated radiotherapy regimen and conventional moderately hypofractionated regimen for localized prostate cancer.

Radiation oncology (London, England)·2026
Same author

MRI-based two-way fluid-structure interaction simulation for discriminating symptomatic carotid atherosclerosis.

Computer methods and programs in biomedicine·2026
Same author

PDAFormer 3+: A full-scale connected modified transformer with parallel dual attention for 3D medical image segmentation.

Artificial intelligence in medicine·2026
Same author

The gp38 protein inhibits host adsorption of phage vB_EcoM_SD286.

Frontiers in microbiology·2026
Same author

Editorial for "Subregional Radiomics Analysis on Multiparametric MRI for Evaluating Lymphovascular Invasion and Survival in Gastric Cancer: A Multicenter Study".

Journal of magnetic resonance imaging : JMRI·2026

Related Experiment Video

Updated: Jul 6, 2025

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
04:23

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images

Published on: April 21, 2023

1.9K

Shape-margin knowledge augmented network for thyroid nodule segmentation and diagnosis.

Weihua Liu1, Chaochao Lin2, Duanduan Chen3

  • 1School of Medical Technology, Beijing Institute of Technology, 5 Zhongguancun South Street, Haidian, 100081, Beijing, China; AthenaEyesCO., LTD., Building 14, No. 39 Jianshan Road, Changsha, 410205, Hunan, China.

Computer Methods and Programs in Biomedicine
|January 9, 2024
PubMed
Summary

This study introduces SkaNet, a novel deep learning model that integrates thyroid nodule segmentation and diagnosis. SkaNet enhances diagnostic accuracy by analyzing shape and margin characteristics, improving computer-aided diagnosis systems.

Keywords:
Knowledge augmented learningMulti-task learningThyroid nodule diagnosisThyroid nodule segmentationUltrasound image analysis

More Related Videos

Author Spotlight: Integrating Ultrasound Imaging with Biochemical Markers for Thyroid Disease Diagnosis
05:41

Author Spotlight: Integrating Ultrasound Imaging with Biochemical Markers for Thyroid Disease Diagnosis

Published on: February 9, 2024

632
Author Spotlight: A 3D Digital Model for the Diagnosis and Treatment of Pulmonary Nodules
10:26

Author Spotlight: A 3D Digital Model for the Diagnosis and Treatment of Pulmonary Nodules

Published on: May 19, 2023

1.9K

Related Experiment Videos

Last Updated: Jul 6, 2025

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
04:23

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images

Published on: April 21, 2023

1.9K
Author Spotlight: Integrating Ultrasound Imaging with Biochemical Markers for Thyroid Disease Diagnosis
05:41

Author Spotlight: Integrating Ultrasound Imaging with Biochemical Markers for Thyroid Disease Diagnosis

Published on: February 9, 2024

632
Author Spotlight: A 3D Digital Model for the Diagnosis and Treatment of Pulmonary Nodules
10:26

Author Spotlight: A 3D Digital Model for the Diagnosis and Treatment of Pulmonary Nodules

Published on: May 19, 2023

1.9K

Area of Science:

  • Medical Imaging Analysis
  • Artificial Intelligence in Healthcare
  • Oncology Diagnostics

Background:

  • Thyroid nodule segmentation and diagnosis are critical for accurate medical assessments.
  • Current computer-aided diagnosis systems often treat segmentation and diagnosis as separate tasks, leading to potential error accumulation.
  • The Thyroid Imaging Reporting and Data System (TI-RADS) highlights the importance of shape and margin characteristics in differentiating thyroid nodules.

Purpose of the Study:

  • To develop a unified framework that integrates thyroid nodule segmentation and diagnosis.
  • To leverage TI-RADS insights by incorporating shape and margin characteristics into a joint learning process.
  • To improve the accuracy and interpretability of thyroid nodule analysis in computer-aided diagnosis.

Main Methods:

  • Proposed SkaNet, a shape-margin knowledge augmented network for simultaneous segmentation and diagnosis.
  • Employed a dual-branch architecture sharing features for both tasks, combining convolutional and self-attention maps.
  • Introduced an exponential mixture module for enhanced discriminative features and a knowledge-augmented multi-task loss with a constraint penalty term embedding shape and margin characteristics.

Main Results:

  • Evaluated SkaNet on public (DDTI) and local thyroid ultrasound datasets.
  • Demonstrated significant improvements in performance compared to state-of-the-art methods.
  • Validated the effectiveness of integrating segmentation and diagnosis with knowledge augmentation.

Conclusions:

  • SkaNet successfully combines thyroid nodule segmentation and diagnosis within a unified, knowledge-augmented framework.
  • The model effectively captures key shape and margin characteristics for improved benign vs. malignant discrimination.
  • This approach offers promising advancements for computer-aided diagnosis systems, particularly in joint segmentation and diagnostic tasks.