基于生成对抗网络的医学图像扩充算法
Medical Image Expansion Algorithm Based on Generative Adversarial Networks
于志勇1
作者信息
- 1. 中国石油大学(华东)计算机科学与技术学院 青岛 266580
- 折叠
摘要
生成对抗网络凭借其强大的拟合能力,已经在许多领域展露峥嵘.论文提出了一种基于生成对抗网络的针对肺部CT图像的图像生成方法,该方法融合了自注意力机制的特征提取能力和生成对抗网络对数据分布的拟合能力,同时针对输入向量进行特异性处理,实验结果表明,论文方法生成图像具有较高的质量以及可用性.
Abstract
Generative adversarial networks have achieved significant results in many fields by virtue of their powerful fitting ability.The paper proposes a generative adversarial network-based image generation method for lung CT images,which combines the feature extraction ability of the self-attentive mechanism and the fitting ability of the generative adversarial network to the data distribution,as well as the specificity processing for the input vector,and the experimental results show that the paper method gen-erates images with high quality as well as usability.
关键词
生成对抗网络/CT图像/自编码/数据集扩充Key words
generative adversarial networks/CT images/self-coding/dataset expansion引用本文复制引用
出版年
2024
国家科技期刊平台
NSTL
万方数据