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

相关实验视频

Updated: Jul 9, 2025

Using Retinal Imaging to Study Dementia
09:17

Using Retinal Imaging to Study Dementia

Published on: November 6, 2017

21.7K

多个规模的多重注意网络用于糖尿病视网膜病变分级.

Haiying Xia1, Jie Long1, Shuxiang Song1

  • 1School of Electronic and Information Engineering, Guangxi Normal University, Guilin 541004, People's Republic of China.

Physics in medicine and biology
|November 30, 2023
PubMed
概括

一个新的多级多注意网络 (MMNet) 通过有效捕捉小病变和多种特征来改善糖尿病视网膜病变 (DR) 的自动分级. 这种AI模型提高了DR查的诊断准确性和效率.

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

Interface-Targeted Rational Design Strategies for Enhancing the Thermostability of d-Allulose 3-Epimerase.

Journal of agricultural and food chemistry·2026
Same author

Low-intensity pulsed ultrasound regulates the osteogenic-adipogenic differentiation balance of rat adipose-derived stem cells via the PI3K/AKT-SREBF1 signaling axis.

Stem cells translational medicine·2026
Same author

Prevention of aggression in psychiatric patients in five psychiatric hospitals in China: a best practice implementation project.

JBI evidence implementation·2026
Same author

Proteomics in bone malignancies: from bulk profiling to single-cell and ultra-low-input technologies.

Journal of translational medicine·2026
Same author

Differentiating Mummified Thyroid Nodules From Papillary Thyroid Carcinoma: A Machine Learning Approach Using Multi-Modal Ultrasound Radiomics.

Ultrasound in medicine & biology·2026
Same author

Circularly Polarized Quasimonochromatic High Harmonic Generation.

Physical review letters·2026

科学领域:

  • 眼科医生 眼科 眼科
  • 医疗成像医学成像
  • 人工智能的人工智能

背景情况:

  • 糖尿病视网膜病变 (DR) 的分级对于临床诊断至关重要,但受到类内变异和小病变的挑战.
  • 深度学习模型经常难以保留来自小病变的信息,并处理DR fundus图像中的特征变化.

研究的目的:

  • 为改进糖尿病视网膜病变的自动分级开发一个新的多规模多注意网络 (MMNet).
  • 解决DR检测中捕获小病变和多种特征变异的现有方法的局限性.

主要方法:

  • 提出了一个病变注意模块,将通道和空间注意力结合起来,编码各种病变特征.
  • 引入了一个多尺度的特征融合模块,以增强小损伤区域的学习.
  • 实现了跨层一致性约束损失,以减轻跨多层特征的语义差异.

主要成果:

  • 在EyePACS数据集上,MMNet在多类DR分级方面实现了86.4%的准确性和88.4%的kappa评分.
  • 在Messidor-1数据集上,MMNet报告了98.6%的AUC,95.3%的准确性,92.7%的回忆,95.0%的精度和93.3%的F1-分数用于转介分类.
  • 在两个基准数据集上表现出与最先进的DR分级方法相比的显著改进.
关键词:
糖尿病视网膜病变的分级损伤注意力模块的病变多尺度特征融合模块多尺度特征融合模块

更多相关视频

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

2.8K
Quantification of Diabetes-induced Adherent Leukocytes in Retinal Vasculature
05:54

Quantification of Diabetes-induced Adherent Leukocytes in Retinal Vasculature

Published on: January 24, 2025

288

相关实验视频

Last Updated: Jul 9, 2025

Using Retinal Imaging to Study Dementia
09:17

Using Retinal Imaging to Study Dementia

Published on: November 6, 2017

21.7K
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

2.8K
Quantification of Diabetes-induced Adherent Leukocytes in Retinal Vasculature
05:54

Quantification of Diabetes-induced Adherent Leukocytes in Retinal Vasculature

Published on: January 24, 2025

288

结论:

  • MMNet有效地提高了糖尿病视网膜病变查的诊断效率和准确性.
  • 拟议的网络推进了计算机辅助医学诊断在DR查中的应用.