• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>Asilomar Conference on Signals, Systems and Computers >Physics and data driven models for ultrasound image reconstruction
【24h】

Physics and data driven models for ultrasound image reconstruction

机译：物理和数据驱动的超声图像重建模型

获取原文
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Ultrasound imaging is widely used in medicine, but image quality is often poor. To address this shortcoming, we developed physical model driven approaches for ultrasound beamforming. We have shown that these methods work well, but they are slow and are unlikely to achieve video frame rates in the near term. We also recently introduced a flexible data-based approach for ultrasound beamforming using deep neural networks (DNN). In principal, the performance between these two methods should be fairly similar. We performed this comparison using anechoic cysts, and the outcomes show that ADMIRE still has slightly better contrast, but that the deep networks had the best contrast-to-noise ratio (CNR). The two approaches do perform similarly, but the DNN training should still be refined.
机译：超声成像在医学中被广泛使用，但是图像质量通常很差。为了解决此缺点，我们开发了物理模型驱动的超声束成形方法。我们已经证明了这些方法效果很好，但是它们很慢并且不太可能在短期内实现视频帧速率。我们最近还推出了一种基于数据的灵活方法，用于使用深度神经网络（DNN）进行超声波束成形。原则上，这两种方法之间的性能应相当相似。我们使用无回声囊肿进行了比较，结果表明ADMIRE的对比度仍然略好，但深层网络具有最佳的对比度-噪声比（CNR）。两种方法的执行效果相似，但DNN训练仍应完善。

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