多级水脊线算法在二维医学超声图像中的应用

Hierarchical Watershed Algorithm’s Application to 2-D Medical Ultrasound Images

以肝囊肿超声图像为例,介绍一种新的医学超声图像的分割方法:多级水脊线分割算法在医学二维超声图像分割中的应用。由于斑点噪声和超声衰减产生的假象,二维和三维超声成像中的图像分割问题一直是公认的一大难点。本文采用由若干数学形态学辅助的水脊线方法提取超声图像中的肝囊肿边缘,取得了相当令人满意的结果,这为进一步的图像处理和化疗等治疗手段奠定了必要的基础。该方法自动选择并分割图像中的主要目标对象,从而克服了传统水脊线算法中常见的过分割和改进水脊线算法中半自动的缺陷。

Taking the liver cyst ultrasound images as the example, this paper introduces hierarchical watershed segmentation , a novel segmentation of medical ultrasound images, in the medical 2-D ultrasound images. Because of the speckle noise and the ultrasound weaken producing the pseudo morph, the segmentation of 2-D and in 3-D ultrasound images always is a recognition big problem. In this paper, watershed algorithm, basing a number of mathematics morphology assistances, produces the liver cyst accurate contours in the ultrasound image. This result benefits for further treatment algorithm on images' processing and the essential foundation for chemotherapy. The proposed scheme automatically selected and finished segmentation essential target object in ultrasound images, and has overcame the common over-segmentation in the traditional watershed algorithm and the semi-automatic flaw in the improving watershed algorithm.