基于深度神经网络的医学图像分类

IF 0.8 4区物理与天体物理 Q4 OPTICS

Optics and Spectroscopy Pub Date : 2025-04-11 DOI:10.1134/S0030400X25700018

D. V. Lvov, A. N. Khaibullin, R. R. Zagidullin, S. A. Voinash

引用次数: 0

摘要

研究结果表明，将ResNet-50架构与预训练的BERT语言模型相结合的模型达到了以下指标：准确率0.94,AUROC 0.87, F1 - score 0.90。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

查看原文本刊更多论文

Classification of Medical Images by Deep Neural Networks

As a result of the research, the model combining the ResNet-50 architecture and the pre-trained BERT language model that has achieved the following indicators: Accuracy 0.94, AUROC 0.87, and F1‑score 0.90 has been developed.

求助全文

通过发布文献求助，成功后即可免费获取论文全文。去求助

来源期刊

Optics and Spectroscopy 物理-光谱学

CiteScore

1.60

自引率

0.00%

发文量

审稿时长

4.5 months

期刊介绍： Optics and Spectroscopy (Optika i spektroskopiya), founded in 1956, presents original and review papers in various fields of modern optics and spectroscopy in the entire wavelength range from radio waves to X-rays. Topics covered include problems of theoretical and experimental spectroscopy of atoms, molecules, and condensed state, lasers and the interaction of laser radiation with matter, physical and geometrical optics, holography, and physical principles of optical instrument making.