拟合正态分布曲线的肺野图像分割与三维重建

摘要

为解决大多数的医学图像分割算法不能完全摆脱人工辅助选取阈值的问题,提出一种基于直方图拟合正态分布曲线寻找种子点的动态分割算法.从分析正态分布曲线如何影响事件发生概率的原理入手,介绍如何采用拟合算法对肺野图像进行分割和提取,利用视觉化工具函式库(VTK)和美国国家卫生院下属国立图书馆开发的医学图像分割与配准算法研发平台(ITK),对分割后的肺野图像进行体绘制三维重建,达到自动寻找种子点、对病变部分进行完整分割和三维可视化的目的.临床图像实验结果表明,与区域生长算法和分水岭算法相比,拟合算法在分割精度和自动化程度上具有显著优势,能够弥补以往常规CT和MR检查手段的不足,增加医生及时、敏锐发现比较小的、早期的和位于隐蔽部分的病灶的机率,为患者设计最佳个性化治疗方案提供了新的可能.%Most of the current image segmentation algorithms cannot be separated from the manually assistant set threshold, such as common region growth algorithm and watershed algorithms.In view of this problem, an dynamical segmentation algorithm was given in which the use of histogram fitting the normal distribution curve to find seed point was proposed.The use of the fitting algorithm for image segmentation and extraction of lung field was introduced, and visualization toolkit(VTK) and insight segmentation and registration toolkit(ITK) were used to develop environment of the lung field segmentation into 3D reconstruction, finding the seed point automatically, segmenting the ill-parts completely and visualizing the medical images.Results of clinical image experiments show that, compared with the region growing algorithm and the watershed algorithm, the proposed method has significant advantages in segmentation precision and automation.It increases chances for the doctors to find small and hidden lesion timely, which provides a way for generating optimal individualized treatment plan.