Jove
Visualize
联系我们
JoVE
x logofacebook logolinkedin logoyoutube logo
关于 JoVE
概览领导团队博客JoVE 帮助中心
作者
出版流程编辑委员会范围与政策同行评审常见问题投稿
图书馆员
用户评价订阅访问资源图书馆顾问委员会常见问题
研究
JoVE JournalMethods CollectionsJoVE Encyclopedia of Experiments存档
教育
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab Manual教师资源中心教师网站
使用条款与条件
隐私政策
政策

相关概念视频

您也可能阅读

相关文章

通过共同作者、期刊和引用图与本文相关的文章。

排序
Same author

A paradigm of hybrid-supervision for annotation-scarce periapical film analysis.

BMC oral health·2026
Same author

A Cytosine‑Bulged (3+1) Hybrid G‑Quadruplex Formed by the Chicken DNA Replication Origin.

Chemistry (Weinheim an der Bergstrasse, Germany)·2026
Same author

A Meta-Analysis Comparing Accelerometer-Based Portable Navigation Systems and Traditional Techniques in Total Knee Arthroplasty: Assessment of Radiological and Clinical Outcomes.

The Journal of arthroplasty·2026
Same author

Fluorinated Hypercrosslinked Polymers with Exceptional Capacity for Perfluorooctanoic Acid Removal from Water.

Small (Weinheim an der Bergstrasse, Germany)·2026
Same author

Deep learning-driven intraoperative assessment of pulp stumps for precision pulpotomy.

Journal of dentistry·2026
Same author

Novel Hinokitiol-Based Ester/Sulfonate Derivatives Containing Diverse Heterocycles as Potential Fungicides and Nematicides.

Chemistry & biodiversity·2026

相关实验视频

Updated: Jan 15, 2026

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

3.3K

自主监督学习增强了带有有限标记数据的周边膜细分.

Meiyu Hu1, Qianli Zhang2, Zhenyang Wei1

  • 1School of Computer and Communication Engineering, University of Science and Technology Beijing, No. 30 Academy Road, Haidian District, Beijing 100083, China.

Journal of dentistry
|October 9, 2025
PubMed
概括

这项研究引入了一个自我监督的学习框架,用于准确的牙周膜细分,大大减少了大量手动注释的需要. 这种方法提高了诊断工具的开发和牙科工作流程的效率.

关键词:
计算机视觉 计算机视觉 计算机视觉深度学习是一种深度学习.数字成像技术的数字成像.多结构细分化的多结构细分.自主监督学习学习

更多相关视频

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

3.2K
Guided Endodontics: Three-Dimensional Planning and Template-Aided Preparation of Endodontic Access Cavities
07:14

Guided Endodontics: Three-Dimensional Planning and Template-Aided Preparation of Endodontic Access Cavities

Published on: May 24, 2022

5.1K

相关实验视频

Last Updated: Jan 15, 2026

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

3.3K
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

3.2K
Guided Endodontics: Three-Dimensional Planning and Template-Aided Preparation of Endodontic Access Cavities
07:14

Guided Endodontics: Three-Dimensional Planning and Template-Aided Preparation of Endodontic Access Cavities

Published on: May 24, 2022

5.1K

科学领域:

  • 医疗成像医学成像
  • 人工智能的人工智能
  • 计算机视觉 计算机视觉

背景情况:

  • 准确细分周周牙膜对于诊断至关重要,但依赖于昂贵的大型标记数据集.
  • 手动细分中的注释变化为AI模型开发引入了不一致性和挑战.

研究的目的:

  • 开发一个自主监督的学习框架,用于周边膜细分,最大限度地减少对广泛标记数据的依赖.
  • 通过减少手动注释努力,提高人工智能在牙科中的实际应用性.

主要方法:

  • 一个两阶段的框架,涉及使用DINOv2重量在未标记的周围膜上进行视觉转换器 (ViT) 的自我监督预训.
  • 在74,292部未标记的周围膜片上,使用学生-教师对比学习进行进一步的预训练.
  • 用Mask2Former头对仅229个标记的片进行细调,以对7个牙结构进行细分.

主要成果:

  • 与传统的监督模型相比 (33.53%-41.55%),自我监督的方法实现了显著更高的子系数 (74.77%).
  • 基于DINOv2的方法超过了其他最先进的自我监督学习方法,包括MAE,MoCov3和BEiTv3.
  • 该方法在统计学上显著优于其监督的Mask2Former同行 (p < 0.01).

结论:

  • 拟议的两阶段,特定于领域的自我监督框架有效地学习了强大的解剖特征,以实现精确的周围膜细分和最小的注释.
  • 这种方法解决了医疗成像中的有限标记数据的挑战,并为人工智能辅助诊断工具提供了可行的途径.
  • 它有望通过减少牙科诊所手动分析时间来提高诊断准确度和提高工作流程效率.