产前超声医学图像处理技术综述

Review of automated medical image analysis in prenatal ultrasound

超声诊断是产科临床上应用最广泛的医学成像方式,与之相应的自动医学图像处理是提高诊断准确率与客观性的重要手段。本文首先介绍超声医学图像处理方法的原理及特点,讨论了若干技术及其涉及的算法,包括图像滤波、图像分割及机器学习技术等,浅析了设计可靠的超声医学图像处理方法的要点。其次,以产前超声医学为背景介绍了这些技术的应用,主要包括标准切面自动提取和生物学参数自动测量。最后,讨论了产前超声智能化诊断的发展方向。

Diagnostic ultrasound is the most widely used medical imaging modality of obstetrics. The corresponding medical image analysis techniques are important for the improvement of diagnostic accuracy and objectivity. This paper first introduces the principles and characteristics of methods in ultrasound medical image analysis, and discusses a number of techniques and corresponding algorithms including image filtering, image segmentation and machine learning. The key points of designing reliable methods for ultrasound medical image analysis are provided. Then, the applications of these techniques, including standard plane selection and biometric measurement, are introduced in the background of obstetrics. Finally, the development direction of intelligent diagnosis in prenatal ultrasound is discussed.