Institute of Biomaterials and Biomedical Engineering, Department of Electrical Engineering, University of Toronto, Toronto, Ontario, Canada.
Ultrasonics. 2013 Feb;53(2):335-44. doi: 10.1016/j.ultras.2012.06.016. Epub 2012 Jul 10.
A technique is proposed for the detection of abnormalities (targets) in ultrasound images using little or no a priori information and requiring little operator intervention. The scheme is a combination of the CLEAN algorithm, originally proposed for radio astronomy, and constant false alarm rate (CFAR) processing, as developed for use in radar systems. The CLEAN algorithm identifies areas in the ultrasound image that stand out above a threshold in relation to the background; CFAR techniques allow for an adaptive, semi-automated, selection of the threshold. Neither appears to have been previously used for target detection in ultrasound images and never together in any context. As a first step towards assessing the potential of this method we used a widely used method of simulating B-mode images (Field II). We assumed the use of a 256 element linear array operating at 3.0MHz into a water-like medium containing a density of point scatterers sufficient to simulate a background of fully developed speckle. Spherical targets with diameters ranging from 0.25 to 6.0mm and contrasts ranging from 0 to 12dB relative to the background were used as test objects. Using a contrast-detail analysis, the probability of detection curves indicate these targets can be consistently detected within a speckle background. Our results indicate that the method has considerable promise for the semi-automated detection of abnormalities with diameters greater than a few millimeters, depending on the contrast.
提出了一种利用很少或没有先验信息且需要很少操作者干预的超声图像异常（目标）检测技术。该方案是 CLEAN 算法和恒虚警率（CFAR）处理的组合，CLEAN 算法最初用于射电天文学，CFAR 技术用于雷达系统。CLEAN 算法识别出与背景相比在超声图像中突出的区域；CFAR 技术允许自适应、半自动地选择阈值。这两种方法似乎都没有以前用于超声图像中的目标检测，也没有在任何情况下一起使用过。作为评估这种方法潜力的第一步，我们使用了一种广泛用于模拟 B 模式图像的方法（Field II）。我们假设使用一个 256 元素的线性阵列，在一个类似于水的介质中以 3.0MHz 的频率工作，该介质中包含足够的点状散射体密度，以模拟完全发展的散斑背景。使用直径从 0.25 毫米到 6.0 毫米且与背景对比度从 0 到 12dB 的球形目标作为测试对象。通过对比细节分析，检测概率曲线表明，在散斑背景下，可以一致地检测到这些目标。我们的结果表明，该方法在一定对比度下，具有很大的潜力用于直径大于几毫米的异常半自动检测。
