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

Cerebral Edema ll: Pathophysiology01:22

Cerebral Edema ll: Pathophysiology

Vasogenic edema is a major form of cerebral edema characterized by abnormal accumulation of fluid in the brain’s extracellular space due to disruption of the blood–brain barrier (BBB). The BBB is a specialized structure composed of endothelial cells connected by tight junctions, supported by astrocytic endfeet and a basement membrane. Under normal conditions, it tightly regulates the movement of ions, proteins, and solutes between the bloodstream and brain parenchyma. When this barrier loses...

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

Surgical Video Understanding with Alignment-Preserving Temporal Adaptation and Action Triplet Text Alignment.

Bioengineering (Basel, Switzerland)·2026
Same author

Rapid immunohistochemistry for intraoperative differentiation between high-grade gliomas and primary central nervous system lymphomas: a multicenter prospective study.

Brain tumor pathology·2026
Same author

Author Correction: Attenuated fusogenicity and pathogenicity of SARS-CoV-2 Omicron variant.

Nature·2026
Same author

Therapeutic Courses of Presumed Intracranial Immature Teratoma Diagnosed Based on Increased Alpha-Fetoprotein.

Journal of child neurology·2026
Same author

Virological characteristics of SARS-CoV-2-related coronaviruses dynamically circulating in Southeast Asia.

Cell·2026
Same author

Extraskeletal Myxoid Chondrosarcoma Mimicking Myoepithelial Tumor.

Case reports in pathology·2026

相关实验视频

Updated: Jun 13, 2026

Brain Infarct Segmentation and Registration on MRI or CT for Lesion-symptom Mapping
10:25

Brain Infarct Segmentation and Registration on MRI or CT for Lesion-symptom Mapping

Published on: September 25, 2019

49.5K

病理基础模型的转移学习策略:脑瘤分类中的系统评估.

Ken Enda1, Yoshitaka Oda1, Zen-Ichi Tanei2

  • 1Department of Cancer Pathology, Faculty of Medicine, Hokkaido University, Sapporo, Japan.

Pathology international
|February 18, 2026
PubMed
概括

对于大脑瘤分类AI,使用基础模型进行线性探测 (LP) 优于对外部数据进行微调 (FT). 这种方法增强了AI的概括性,克服了有限的医院病例数据的局限性.

更多相关视频

Modeling Brain Metastases Through Intracranial Injection and Magnetic Resonance Imaging
06:44

Modeling Brain Metastases Through Intracranial Injection and Magnetic Resonance Imaging

Published on: June 7, 2020

8.1K
Quantifying the Brain Metastatic Tumor Micro-Environment using an Organ-On-A Chip 3D Model, Machine Learning, and Confocal Tomography
09:53

Quantifying the Brain Metastatic Tumor Micro-Environment using an Organ-On-A Chip 3D Model, Machine Learning, and Confocal Tomography

Published on: August 16, 2020

7.8K

相关实验视频

Last Updated: Jun 13, 2026

Brain Infarct Segmentation and Registration on MRI or CT for Lesion-symptom Mapping
10:25

Brain Infarct Segmentation and Registration on MRI or CT for Lesion-symptom Mapping

Published on: September 25, 2019

49.5K
Modeling Brain Metastases Through Intracranial Injection and Magnetic Resonance Imaging
06:44

Modeling Brain Metastases Through Intracranial Injection and Magnetic Resonance Imaging

Published on: June 7, 2020

8.1K
Quantifying the Brain Metastatic Tumor Micro-Environment using an Organ-On-A Chip 3D Model, Machine Learning, and Confocal Tomography
09:53

Quantifying the Brain Metastatic Tumor Micro-Environment using an Organ-On-A Chip 3D Model, Machine Learning, and Confocal Tomography

Published on: August 16, 2020

7.8K

科学领域:

  • 人工智能在病理学中的应用
  • 计算病理学计算病理学
  • 机器学习用于医学成像

背景情况:

  • 在医院部署病理学AI受到有限的病例数据和适应预先训练的模型的挑战.
  • 脑瘤的分类是一个特殊的挑战,因为不同的类别和少数机构案例.
  • 在这种情况下,基础模型的最佳转移学习策略尚不清楚.

研究的目的:

  • 用基础和传统AI模型评估微调 (FT) 与线性探测 (LP) 进行脑瘤分类.
  • 确定最有效的转移学习策略,以便将AI模型适应本地和外部数据集.
  • 评估不同转移学习方法对AI模型概括能力的影响.

主要方法:

  • 对基础模型 (UNI,Prov-GigaPath) 和传统模型 (ViT-L,CTransPath) 的FT和LP策略进行了比较.
  • 在机构数据集 (254个案例) 上训练模型,并在EBRAINS数据集 (698个案例) 上验证.
  • 基于对数据集的分类准确性和概括性来评估性能.

主要成果:

  • 传统模型显示FT ≥LP性能在两个数据集.
  • 基金会模型呈现逆转:FT在机构数据上略高于LP,而LP在外部数据上明显优于FT (p < 0.01).
  • 在外部验证 (p < 0.001) 中,使用LP (10个补丁/盒) 的UNI超过了微调的常规模型 (500个补丁/盒).

结论:

  • 对有限的机构数据进行微调的基础模型可能会导致过度调整和妥协概括.
  • 线性探测保留了预先训练的表示,使人工智能实现更有效,并为脑瘤分类提供了优越的概括.
  • 线性探测成为部署基础模型在病理学AI中的更强大的战略,局部数据有限.