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

Issues And Trends In Healthcare Delivery System01:29

Issues And Trends In Healthcare Delivery System

5.8K
The issues and trends in healthcare delivery are constantly changing. The COVID-19 pandemic is one recent issue that wreaked havoc on healthcare systems, causing a shortage of healthcare workers, high demand for medicines and supplies, and increased medical expenditure due to a lack of insurance. Other issues include rising healthcare costs and care fragmentation.
Cost Containment
Payment for healthcare services has historically promoted adoption of costly and often unnecessary or inefficient...
5.8K

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

Effect of risperidone treatment on insulin-like growth factor-1 and interleukin-17 in drug naïve first-episode schizophrenia.

Psychiatry research·2021
Same author

Circularly polarized luminescent systems fabricated by Tröger's base derivatives through two different strategies.

Beilstein journal of organic chemistry·2021
Same author

Cerebral Infarction in Immune Thrombotic Thrombocytopenic Purpura Is Associated with Old Age, Hypertension, Smoking, and Anti-ADAMTS13 Ig, But Not with Mortality.

TH open : companion journal to thrombosis and haemostasis·2021
Same author

<sup>1</sup>H, <sup>13</sup>C, and <sup>15</sup>N resonance assignments of reduced apo-WhiB4 from Mycobacterium tuberculosis.

Biomolecular NMR assignments·2021
Same author

The mitochondrial genome of <i>Suillia</i> sp. (Diptera: Heleomyzidae).

Mitochondrial DNA. Part B, Resources·2020
Same author

The mitochondrial genome of <i>Medetera</i> sp. (Diptera: Dolichopodidae).

Mitochondrial DNA. Part B, Resources·2020

相关实验视频

Updated: Sep 9, 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

2.9K

基于扩散模型的医疗图像生成作为AI应用的潜在数据增强策略

Zijian Cao1, Jueye Zhang2, Chen Lin2

  • 1Institute of Medical Technology, Peking University Health Science Center, Beijing 100191, China.

Current medical imaging
|September 5, 2025
PubMed
概括

这项研究引入了一种用于生成合成医疗图像的扩散模型,为人工智能 (AI) 培训提供了一种高效的数据增强方法. 这项研究突出了高质量的合成数据生成的最佳参数,即使资源有限.

关键词:
人工智能培训人工智能数据增强扩散模型图像生成医学放射学

更多相关视频

Author Spotlight: Advancing CBCT and Digital Dental Image Integration with AI-Assisted Digitization
05:49

Author Spotlight: Advancing CBCT and Digital Dental Image Integration with AI-Assisted Digitization

Published on: February 23, 2024

985
Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness
03:14

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Published on: December 6, 2024

680

相关实验视频

Last Updated: Sep 9, 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

2.9K
Author Spotlight: Advancing CBCT and Digital Dental Image Integration with AI-Assisted Digitization
05:49

Author Spotlight: Advancing CBCT and Digital Dental Image Integration with AI-Assisted Digitization

Published on: February 23, 2024

985
Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness
03:14

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Published on: December 6, 2024

680

科学领域:

  • 医学成像
  • 人工智能
  • 计算机视觉

背景情况:

  • 医疗人工智能 (AI) 应用需要大量多样化的数据集来进行强有力的培训.
  • 由于隐私问题和成本,对现实世界的临床数据的访问可能受到限制.
  • 数据增强策略对于提高有限数据集的人工智能模型性能至关重要.

研究的目的:

  • 探索使用扩散模型增强医学数据的生成图像合成方法.
  • 在低资源计算环境中评估扩散模型的效率和成本效益.
  • 确定高准确度合成医学图像生成的最佳训练参数.

主要方法:

  • 在低性能计算条件下使用MedMNIST v2数据集进行训练.
  • 根据现有数据特征开发了一个注释扩散模型来合成新的医学图像.
  • 使用不同损失函数和特征向量尺寸的损失函数梯度下降和Fréchet初始距离 (FID) 进行定量评估.

主要成果:

  • 与原始数据相比,扩散模型成功生成了具有类似风格但不同解剖细节的医学图像.
  • 具有特征向量维度为 64 的 L2 损失函数得到了 0.85 的最佳 FID 分数.
  • 胡贝尔损失函数显示了增强的模型稳定性,尽管在特征向量维度为2048时,FID高于15.2.

结论:

  • 基于扩散模型的医学图像合成是人工智能的可行增强策略,特别是当真实数据稀缺时.
  • 最佳的训练参数,包括损失函数选择和特征向量维度,显著影响合成图像质量.
  • 进一步的研究应侧重于将这些模型应用于更复杂的医疗数据集和临床场景.