基于加权霍夫变换的三维超声胎儿脑中线的自动提取
首发时间:2012-07-02
摘要:胎儿脑中线的自动提取是实现胎儿小脑自动分割的必要步骤之一,同时也是诊断胎儿先天性脑中线异常的重要依据。由于胎儿头部结构复杂,且超声图像中斑点噪声分布较多、边缘模糊,使得胎儿脑中线的自动提取存在一定难度。霍夫变换是检测直线的有效方法,但是其准确性容易受到噪声的影响,因此在超声图像中的应用受到噪声的约束。本文提出一种基于高斯模板的加权霍夫变换算法,可以提高算法的鲁棒性,实现在超声图像上自动提取胎儿脑中线。我们将本文算法提取的结果与两位专家手动提取的结果进行准确性分析和T检验,实验结果表明,两种方法提取的结果并没有显著差异,且自动提取的误差较小。因此本文算法是一种有效地自动检测胎儿脑中线的方法。
关键词: 生物医学电子学 三维超声图像 直线提取 胎儿脑中线 加权霍夫变换
For information in English, please click here
Automated detection of fetal brain midline in 3-D ultrasound images using weighted Hough transform
Abstract:The automatic detection of fetal brain midline in 3-D ultrasound images is an essential step in the segmentation of fetal cerebellum, and can provide parameters for prenatal diagnosis of congenital midline abnormalities. However, the complex structures of head and speckle noise complicate the detection without manual intervention. The Hough transform algorithm can effectively detect lines in the images, but it shows poor robustness to noise which is an non-ignorable factor for ultrasound images. In this paper, we proposed a weighted Hough transform (WHT) algorithm based on Gaussian mask to automatically detect the fetal brain midline. We compared the performance of our methods with that of manual segmentations performed by two specialists. The results showed that the WHT produced very similar results to those based on manual tracing.
Keywords: biomedical electronics 3-D ultrasound line extraction fetal brain midline weighted Hough transform
论文图表:
引用
No.****
同行评议
勘误表
基于加权霍夫变换的三维超声胎儿脑中线的自动提取
评论
全部评论0/1000