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

Genital mycoplasma infection: a systematic review and meta-analysis.

Reproductive health·2023
Same author

Molecularly designed and synthesized of bright blue nitrogen-doped lignin-derived carbon dots applied in printable anti-counterfeiting.

International journal of biological macromolecules·2023
Same author

Children's use of reasoning by exclusion to infer objects' identities in working memory.

Journal of experimental child psychology·2023
Same author

The osmotic stress of Vallisneria natans (Lour.) Hara leaves originating from the disruption of calcium and potassium homeostasis caused by MC-LR.

Water research·2023
Same author

Clinical Study of Percutaneous Endoscopic Large-channel Fusion and Transforaminal Lumbar Interbody Fusion in the Treatment of Degenerative Lumbar Spinal Stenosis.

Alternative therapies in health and medicine·2023
Same author

Pan-cancer analyses reveal GTSE1 as a biomarker for the immunosuppressive tumor microenvironment.

Medicine·2023

相关实验视频

Updated: Jul 5, 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

2.8K

一种基于meta-learning模型的难度意识和任务增强方法,用于几次射击糖尿病视网膜病变的分类.

Xueyao Liu1, Xueyuan Dong1, Tuo Li2

  • 1Shandong Key Laboratory of Medical Physics and Image Processing, Shandong Institute of Industrial Technology for Health Sciences and Precision Medicine, School of Physics and Electronics, Shandong Normal University, Jinan, China.

Quantitative imaging in medicine and surgery
|January 15, 2024
PubMed
概括

一种新的难度感知和任务增强元学习 (DaTa-ML) 模型提高了糖尿病视网膜病变 (DR) 分类准确性,但数据有限. 这种方法提高了早期的DR诊断,通过显著减少训练时间和数据,优于现有技术.

关键词:
糖尿病视网膜病变的分类 (DR分类)意识到困难的 (Da)几次射击,几次射击.这就是meta-learning的意义.任务增加 (Ta)

更多相关视频

A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
12:18

A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment

Published on: January 11, 2020

7.5K
Development of a Gaze-Contingent Display Framework Designed for Perceptual and Oculomotor Research with Simulated Central Vision Loss
07:12

Development of a Gaze-Contingent Display Framework Designed for Perceptual and Oculomotor Research with Simulated Central Vision Loss

Published on: April 11, 2025

358

相关实验视频

Last Updated: Jul 5, 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

2.8K
A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
12:18

A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment

Published on: January 11, 2020

7.5K
Development of a Gaze-Contingent Display Framework Designed for Perceptual and Oculomotor Research with Simulated Central Vision Loss
07:12

Development of a Gaze-Contingent Display Framework Designed for Perceptual and Oculomotor Research with Simulated Central Vision Loss

Published on: April 11, 2025

358

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:

  • 准确的糖尿病视网膜病变 (DR) 分类对于早期诊断和治疗至关重要.
  • 有限的注释DR数据挑战了当前的深度学习模型.
  • 对于DR分类,需要采用近距离学习方法.

研究的目的:

  • 提出一种基于元学习 (DaTa-ML) 的难度意识和任务增强方法,用于短暂的DR分类.
  • 为了应对基金图像中有限的注释DR数据的挑战.
  • 提高DR分类模型的效率和准确性.

主要方法:

  • 实施了难度意识 (Da) 方法,以动态调整交叉损失,优先考虑具有挑战性的任务.
  • 利用任务增强 (Ta) 方法通过图像旋转增加元训练任务,增强特征提取.
  • 优化了元培训任务采样和初始化参数,以改善元概括.

主要成果:

  • 在APTOS 2019数据集上,DaTa-ML模型在仅使用1%的训练数据 (5-way,20-shot) 和单个更新步骤时,实现了89.6%的准确性.
  • 与转移学习 (ResNet50在ImageNet上预训练) 相比,表现出1.7%的改进,与从头构建的模型相比,表现出16.8%的改进.
  • 仅在ResNet50参数的0.47%的情况下获得了这些结果,这表明效率很高.

结论:

  • 达塔-ML模型为DR分类提供了一个高效的解决方案,使用最小的注释数据.
  • 该模型在少数射击学习场景中显示了与最先进的方法相比的显著优势.
  • DaTa-ML可以帮助眼科医生确定DR的严重程度,改善患者的护理.