首页> 外文会议>Chinese Intelligent Systems Conference >Generating Synthesized CT from Cone-Beam Computed Tomography (CBCT) Using Artifact Disentanglement Network for Image-Guided Radiotherapy (IGRT)
【24h】

Generating Synthesized CT from Cone-Beam Computed Tomography (CBCT) Using Artifact Disentanglement Network for Image-Guided Radiotherapy (IGRT)

机译:使用工件解剖学网络从锥形光束计算机断层扫描网络(IGRT)产生合成的CT。用于图像引导放射疗法(IGRT)

获取原文

摘要

Recently, image generation, including generating synthesized CT (sCT) from CBCT, has become a research hotspot in medical image analysis. Many sCT generation architectures learn the mapping from CBCT images to corresponding CT images based on an unpaired image-to-image translation method, named cycle-consistent generation adversarial network (CycleGAN), but the anatomy accuracy of the sCT images generated by them needs to be improved. To address this problem, we propose a novel CBCT-based sCT generation architecture, which generates sCT images with CBCT anatomy and CT image quality and is trained with unpaired data. The sCT images generated by the proposed architecture have impressively high anatomy accuracy. Experiments show that the sCT images generated by the proposed method are more visually and quantitatively similar to real CT images than some recently-developed CycleGAN-based sCT generation architectures with the mean peak-signal-to-noise ratio (PSNR) improved from 26.15 dB to 29.66 dB.
机译:最近,来自CBCT的图像生成包括生成合成的CT(SCT),已成为医学图像分析的研究热点。许多SCT一代架构基于未配对的图像到图像转换方法,以命名的循环 - 一致生成的对手网络(CniCalgan)来学习从CBCT图像到对应的CT图像映射,但它们需要产生的SCT图像的解剖学精度改善。为了解决这个问题,我们提出了一种基于CBCT的基于CBCT的SCT生成架构,其使用CBCT解剖和CT图像质量生成SCT图像,并用未配对数据训练。由拟议的体系结构产生的SCT图像令人印象深刻地解剖学准确性高。实验表明,由所提出的方法生成的SCT图像比其他最近开发的基于Cyclegan的SCT的SCT产生架构更具视觉和定量地与具有平均峰值信噪比(PSNR)的最近开发的Cyclan-噪声比(PSNR)从26.15dB改善到29.66 dB。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号