基于深度学习的超声成像技术研究现状

超声成像技术以其无损、无辐射、实时性好、成像深度深和检查费用低等优点，是目前临床医学最常应用的影像技术之一。随着医疗技术的日益发展，兼顾高帧率和高质量的超声成像已成为临床上的迫切需求。深度学习作为能快速提取信号特征的技术，近年来已经在超声成像领域展开多种应用研究并产生了较好的效果，具有很大的应用前景。文章总结了基于深度学习的超声成像技术现状，重点介绍了超声成像技术中常用深度学习架构，列举了深度学习技术在超声成像中的应用，最后总结了目前深度学习在训练以及实现临床应用中可能会遇到的挑战。

Research status of deep learning based ultrasonic imaging

Ultrasonic imaging is one of the most often used clinical imaging and diagnostic methods because of its moderate, real-time, well penetrative and cost-efficient performances. In recent years, the high frame-rate and high quality ultrasonic imaging method is urgently demanded. Meanwhile, as a popular tool to acquire signal features in a short period, deep learning has made impressive improvements and shown great potential in ultrasonic imaging. In this paper, the deep learning method and its applications related to ultrasonic imaging are reviewed, and challenges in deep learning training and clinical practices are also summarized.