Search research articles

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 Journal Methods Collections JoVE Encyclopedia of Experiments 存档

教育

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual 教师资源中心教师网站

使用条款与条件

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

Development and Evaluation of a Keypoint-Based Video Stabilization Pipeline for Oral Capillaroscopy.

Sensors (Basel, Switzerland)·2025

Same author

Automated Stabilization, Enhancement and Capillaries Segmentation in Videocapillaroscopy.

Sensors (Basel, Switzerland)·2023

Same author

NeuronAlg: An Innovative Neuronal Computational Model for Immunofluorescence Image Segmentation.

Sensors (Basel, Switzerland)·2023

Same author

Window-Based Energy Selecting X-ray Imaging and Charge Sharing in Cadmium Zinc Telluride Linear Array Detectors for Contaminant Detection.

Sensors (Basel, Switzerland)·2023

Same author

Evaluation of the Oral Microcirculation in Patients Undergoing Anti COVID-19 Vaccination: A Preliminary Study.

Vaccines·2022

Same author

Recognizing the Emergent and Submerged Iceberg of the Celiac Disease: ITAMA Project-Global Strategy Protocol.

Pediatric reports·2022

Search research articles

相关实验视频

Updated: Jan 7, 2026

Author Spotlight: Impact of Intergenic Interactions on Disease-Identifying Dark Biomarkers

Author Spotlight: Impact of Intergenic Interactions on Disease-Identifying Dark Biomarkers

Published on: March 1, 2024

具有特征的转移学习方法通过高不平衡和多标签分类的基因算法进行区块选择 HPA同焦显微镜图像的基因算法.

Vincenzo Taormina¹, Domenico Tegolo¹, Cesare Valenti¹

¹Dipartimento di Matematica e Informatica, Università degli Studi di Palermo, 90123 Palermo, Italy.

Bioengineering (Basel, Switzerland)

|December 30, 2025

概括

这项研究引入了一种新的两相转移学习方法,用于分析人类蛋白质图谱 (HPA) 数据集中的复杂医学图像. 该方法通过整合图像和细胞水平分析来增强罕见蛋白质模式的识别,实现高F1得分.

关键词:

二元相关性二元相关性一个共聚焦显微镜.卷积神经网络是一种卷积神经网络.光成像的图像光图像遗传算法是一种遗传算法.图片图片图片图片图片标签权力设置标签权力多类多标签分类的分类.支持矢量机器的支持矢量机器.转移学习转移学习

更多相关视频

Selecting Multiple Biomarker Subsets with Similarly Effective Binary Classification Performances

Selecting Multiple Biomarker Subsets with Similarly Effective Binary Classification Performances

Published on: October 11, 2018

Predicting Treatment Response to Image-Guided Therapies Using Machine Learning: An Example for Trans-Arterial Treatment of Hepatocellular Carcinoma

Predicting Treatment Response to Image-Guided Therapies Using Machine Learning: An Example for Trans-Arterial Treatment of Hepatocellular Carcinoma

Published on: October 10, 2018

相关实验视频

Last Updated: Jan 7, 2026

Author Spotlight: Impact of Intergenic Interactions on Disease-Identifying Dark Biomarkers

Author Spotlight: Impact of Intergenic Interactions on Disease-Identifying Dark Biomarkers

Published on: March 1, 2024

Selecting Multiple Biomarker Subsets with Similarly Effective Binary Classification Performances

Selecting Multiple Biomarker Subsets with Similarly Effective Binary Classification Performances

Published on: October 11, 2018

Predicting Treatment Response to Image-Guided Therapies Using Machine Learning: An Example for Trans-Arterial Treatment of Hepatocellular Carcinoma

Predicting Treatment Response to Image-Guided Therapies Using Machine Learning: An Example for Trans-Arterial Treatment of Hepatocellular Carcinoma

Published on: October 10, 2018

科学领域:

计算生物学是一种计算生物学.
医学成像分析分析医学成像分析
深度学习应用程序深度学习应用程序

背景情况:

深度学习在图像分类方面表现出色,但在复杂的医学成像数据集中面临挑战,例如人类蛋白质图谱 (HPA).
由于高类不平衡,罕见模式和多标签分类要求,HPA数据集存在计算困难.
该数据集包括28个模式,超过500个标签组合,以及每个样本的四个不同的成像通道 (绿色,红色,蓝色,黄色).

研究的目的:

开发一种有效的深度学习策略,用于分析复杂的人类蛋白质图谱数据集.
为了提高细胞结构内的罕见蛋白质局部化模式的准确识别.
利用预训练的卷积神经网络 (CNN) 的转移学习和特征提取来进行增强的分类.

主要方法:

一个双相转移学习方法,使用从12个 ImageNet 预训练的 CNN 提取特征块.
第一阶段:使用CNN作为特征提取器和支持矢量机 (SVM) 分类器进行单标签多类分类,结合特征块选择的遗传算法.
第二阶段:应用两个多标签分类策略,整合图像级 (通道分析) 和细胞级 (核和核膜环特征) 分析.

主要成果:

来自不同CNN的连接特征块在第一阶段显著提高了性能.
图像层面的分析揭示了绿色通道的个人优势以及将其与红色和黄色通道相结合的好处.
聚焦于细胞核和核膜环的细胞水平分析有效地识别了罕见的模式,优于现有方法.

结论:

拟议的两阶段转移学习方法有效地解决了HPA数据集的挑战,特别是罕见模式识别.
图像层面和细胞层面分析的整合提高了微妙和不频繁的蛋白质定位的检测.
该研究在"杆与环"模式中获得了值得注意的F1分数0.8,表明了生物和临床应用的显著进展.