基于Bayes分类算法和活动轮廓模型的新疆肝包虫CT图像分割方法

Liver Hydatid CT Image Segmentation Using Smoothed Bayesian Classification Method and Modified Parametric Active Contour Model

肝包虫病是新疆常见的寄生虫病,严重危及人类健康。目前,医院常采用CT影像技术对该病进行诊断。肝包虫CT图像有其特有的病理特征,图像的灰度分布存在不均匀性和边界模糊性,且不同的包虫囊肿类型,其CT图像表现各异。本文针对该病的CT影像特征,提出一种同时对肝脏及包虫病灶进行分割的迭代算法。在每一步迭代过程中,算法分为初始分割和优化分割两个步骤:首先,在CT切片图像中确定位于正常肝脏及包虫病灶区的种子点,根据种子点的位置,利用Gauss概率模型拟合不同区域的灰度分布,并结合Bayes分类算法对肝脏及病灶区同时进行初始分割;然后,利用基于先验形状力场的活动轮廓模型算法优化初始分割结果,从而获得精确的肝脏及病灶区的边界。为了验证该算法的有效性,将算法对不同病人的CT切片图像进行分割实验,并从主观和客观两个方面,将算法的分割结果与医师手动分割结果进行对比评估,结果表明,该算法能在分割肝脏的同时准确地提取包虫病灶区。

Liver hydatid is a common parasitic disease in Xinjiang and a big concern for peoplels health. At present, CT imaging analysis is always a method for diagnosising liver hydatid. The CT image of liver hydatid owns their characteristics, such as inconsistent gray distribution and fuzzy regional boundary. Meanwhile, the representations of CT images are also dissimilar among different types of liver hydatid cyst. Based on CT imaging features of this disease, an iterative approach for liver segmentation and hydatid lesion extraction is proposed in this paper. Each iteration consists of two main steps. Firstly, according to the user-defined pixel seeds in the liver and lesion which are defined by user, Gaussian probability model fitting is adopted to fit gray distribution in different regions and smoothed Bayesian classification is applied to obtain the initial segmentation results of liver and lesion. Secondly, the parametric active contour model using the priori shape force field is adopted to refine the initial segmentation and to get accurate boundaries of liver and lesion. The algorithm from subjective and objective aspects are evaluated on different patientsI CT slices. By comparing the algorithm of segmentation to the ground-truth manual segmentation. The proposed algorithm is shown to be effective in liver segmentation and hydatid lesion extraction.