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

Clinical Policy: Critical Issues Related to Harms of Cannabis Exposure in Adult Patients Presenting to the Emergency Department, Cardiovascular Considerations: Approved by ACEP Board of Directors September 4, 2025.

Annals of emergency medicine·2025
Same author

Influences on Emergency Clinician Use of Health Information Exchange: Interview Study.

JMIR medical informatics·2025
Same author

Clinical Policy: Critical Issues in the Management of Adult Patients Requiring Endotracheal Intubation in the Emergency Department.

Annals of emergency medicine·2025
Same author

Clinical Policy: A Critical Issue in the Outpatient Management of Adult Patients Presenting to the Emergency Department With Asymptomatic Elevated Blood Pressure: Approved by the ACEP Board of Directors January 22, 2025.

Annals of emergency medicine·2025
Same author

Critical Issues in the Evaluation and Management of Adult Patients With Suspected Acute Nontraumatic Thoracic Aortic Dissection.

Annals of emergency medicine·2025
Same author

A Critical Issue in the Management of Adult Patients Presenting to the Emergency Department With Acute Carbon Monoxide Poisoning: Approved by the ACEP Board of Directors January 22, 2025.

Annals of emergency medicine·2025

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

在临床皮肤病学图像分类器的无监督SoftOtsuNet增强.

Miguel Dominguez1, John T Finnell1

  • 1VisualDx, Rochester, NY.

AMIA ... Annual Symposium proceedings. AMIA Symposium
|January 15, 2024
PubMed
概括

本研究介绍了SoftOtsuNet,这是一种无监督深度神经网络 (DNN) 方法,用于机器学习中的数据增强. 它有效地从临床皮肤病图像中删除不相关的背景信息,提高DNN准确性并减少偏差.

科学领域:

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

背景情况:

  • 数据增强对于机器学习 (ML) 模型准确性至关重要,并减少深度神经网络 (DNN) 的过度匹配.
  • 临床皮肤学图像经常包含不相关的背景细节,这可能会引入DNN的偏见.
  • 由于广泛的标签要求,皮肤病图像中前景/背景的监督细分成本很高.

研究的目的:

  • 开发一种新的无监督深度神经网络 (DNN),用于临床皮肤病学中的数据增强.
  • 为应对皮肤病图像中无关紧要背景信息的挑战,这些图像可能会偏向DNN模型.
  • 提出一个具有成本效益的解决方案,避免用于图像分割的手动标签.

主要方法:

  • 开发了一种无监督的DNN,该DNN包含了Otsu方法的可差异化适应.
  • CutOut增强功能与Otsu的方法适应用于动态掩盖相结合.
  • 拟议的SoftOtsuNet在多个皮肤病图像数据集上进行了评估.

主要成果:

  • 与其他增强方法相比,SoftOtsuNet在三个数据集中表现出卓越的性能.
  • 改善包括0.75%的Fitzpatrick17k,1.76%的多样性皮肤病图像和0.92%的专有数据集.

更多相关视频

Combining Reflectance Confocal Microscopy with Optical Coherence Tomography for Noninvasive Diagnosis of Skin Cancers via Image Acquisition
09:37

Combining Reflectance Confocal Microscopy with Optical Coherence Tomography for Noninvasive Diagnosis of Skin Cancers via Image Acquisition

Published on: August 18, 2022

2.3K
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

2.5K

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
Combining Reflectance Confocal Microscopy with Optical Coherence Tomography for Noninvasive Diagnosis of Skin Cancers via Image Acquisition
09:37

Combining Reflectance Confocal Microscopy with Optical Coherence Tomography for Noninvasive Diagnosis of Skin Cancers via Image Acquisition

Published on: August 18, 2022

2.3K
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

2.5K
  • 该方法只影响培训时间,不影响推断成本.

    • 结论:

    • 无监督,人为设计的损失函数仍然可以增强大型,数据驱动的模型.
    • SoftOtsuNet提供了一种有效和高效的方法,用于医疗成像中的数据增强.
    • 这种方法可以减轻DNN中背景诱导的偏差,而不会增加计算推理成本.