Scholarlink

Cerebrovascular diseases seriously threaten human life and health. Digital subtraction angiography (DSA) is often used for diagnosis and treatment of cerebrovascular diseases. However, the quality of DSA is unsatisfactory due to artifacts. In this paper, we proposed a novel framework based on deep learning with a novel decoupling training strategy to reduce artifacts. This strategy can help us to generate DSA images without mask images. The results showed that our proposed method can generate less or even no artifact DSA images with SSIM of 0.9052. In conclusion, our method can produce high-quality DSA images to help doctors decrease the misdiagnosis rate due to artifacts.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

GDN: Generative Decoupling Network for Digital Subtraction Angiography Generation

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Ruibo Liu,
  • Ronghui Tian,
  • Zhenzhou Li,
  • Ligang Chen,
  • Xinyu Yang,
  • Wei Qian,
  • Guobiao Liang,
  • Guangxin Chu,
  • Hai Jin,
  • He Ma

摘要

Cerebrovascular diseases seriously threaten human life and health. Digital subtraction angiography (DSA) is often used for diagnosis and treatment of cerebrovascular diseases. However, the quality of DSA is unsatisfactory due to artifacts. In this paper, we proposed a novel framework based on deep learning with a novel decoupling training strategy to reduce artifacts. This strategy can help us to generate DSA images without mask images. The results showed that our proposed method can generate less or even no artifact DSA images with SSIM of 0.9052. In conclusion, our method can produce high-quality DSA images to help doctors decrease the misdiagnosis rate due to artifacts.