相关实验视频

Updated: May 2, 2026

A Method for Labeling Vasculature in Embryonic Mice

A Method for Labeling Vasculature in Embryonic Mice

Published on: October 7, 2011

无监督学习为标记全球质硬化症

Hrafn Weishaupt¹, Justinas Besusparis¹, Cleo-Aron Weis²

¹Department of Pathology, Haukeland University Hospital, Bergen, 5021, Norway.

Computers in biology and medicine

|August 2, 2025

概括

聚类质细胞图像对于预标签非常准确,在大多数情况下达到95%以上的准确性. 这支持无监督学习和病分类模型的交互标签.

科学领域:

腎病理學腎病理學
计算病理学计算病理学
医学图像分析医学图像分析

背景情况:

在病理学中,监督学习需要专家进行耗时的手动图像标签.
没有监督的策略,如质细胞聚类,可以减少这种标签瓶.
之前的研究建议对半自动或无监督训练进行分组,但其对基本质细胞分类的准确性尚不清楚.

研究的目的:

评估聚类对分离全球性硬化和非全球性硬化质细胞的准确性和局限性.
为了解决对基本质细胞分类的聚类性能理解的差距.

主要方法:

对10个不同标记的病理学数据集进行了聚类.
从34个不同预训练的卷积神经网络 (CNN) 模型中使用了特征嵌入.

主要成果:

聚类证明了分离全球性和非全球性硬化球体的高可行性.
在大多数测试数据集中,准确度超过95%.

结论:

聚类是一种非常准确的前标记质细胞的方法,作为下游交互式或无监督学习的坚实基础.
这些发现推动了临床上适用的淋巴细胞分类模型的开发.
未来的工作应该通过对比学习或基础模型来探索特定领域的特征提取器,以提高性能.

关键词:

集群集成是指集群集成.深度学习是一种深度学习.它们的质细胞.葡萄细胞硬化症葡萄细胞硬化症腎病理學腎病理學没有监督的学习学习.

更多相关视频

Supervised Machine Learning for Semi-Quantification of Extracellular DNA in Glomerulonephritis

Supervised Machine Learning for Semi-Quantification of Extracellular DNA in Glomerulonephritis

Published on: June 18, 2020

Author Spotlight: Aiding Research in Kidney Biology by Labeling Glomeruli in Cleared Tissues

Author Spotlight: Aiding Research in Kidney Biology by Labeling Glomeruli in Cleared Tissues

Published on: February 9, 2024

相关概念视频

Conduct Disorder

Conduct Disorder

Conduct disorder is a complex mental health diagnosis characterized by a repetitive and persistent pattern of behavior that violates societal norms, the rights of others, or age-appropriate rules. The diagnostic criteria for conduct disorder require the presence of at least three problematic behaviors within the past 12 months, with at least one occurring in the past six months. These behaviors are grouped into four categories: aggression toward people and animals; destruction of property;...