改进GAC模型肺部薄扫CT图像序列分割法  

Improved GAC model for lung thin CT image sequences segmentation algorithm

作　　者：贺娜娜 强彦[1] 赵涓涓[1] 郝晓燕[1] 

机构地区：[1]太原理工大学计算机科学与技术学院,山西太原030024

出　　处：《计算机工程与设计》2017年第10期2772-2777,共6页Computer Engineering and Design

基　　金：国家自然科学基金项目(61373100;61540007;;61572344);国家重点实验室开放基金项目(BUAA-VR-15KF02;BUAA-VR-16KF-13)

摘　　要：针对测地线活动轮廓(geodesic active contour,GAC)模型轮廓演化速度慢的问题,构造一个区域灰度相似性信息项,对GAC模型的能量泛函进行改进,加快轮廓演化速度,将其用于肺部薄扫CT(computed tomography)图像序列中肺实质的自动分割。采用基于Nystrom逼近的谱聚类算法分割CT图像序列中间位置CT中的肺实质,计算其灰度均值与标准差,构造区域灰度相似性信息项,以分割好的肺实质轮廓作为初始轮廓,分别从上下两个方向采用改进了能量泛函的GAC模型实现其它切片中肺实质的分割。实验结果表明,该方法能够较好实现肺实质的自动分割,与医师分割结果的重合率可达94.83%,时间消耗较少。Aiming at the problem that the contour of geodesic active contour（GAC）model evolves slowly,an energy term concerning gray similarity information was constructed to modify the energy function of GAC model,which accelerated the speed of contour evolution.This improved algorithm was used to segment lung parenchyma of thin computed tomography（CT）image sequences.Lung parenchyma of the CT image in the middle of the image sequences was segmented using Nystrom based spectral clustering algorithm.The gray mean and gray standard variation of segmented lung parenchyma were calculated to construct the gray similarity information term.The contour of segmented lung parenchyma was made as initial contour,and the improved GAC model was applied to segment lung parenchyma of other CT slices.Experimental results demonstrate that the proposed algorithm can segment lung parenchyma automatically and achieve an average volume overlap ratio of 94.83% compared with the segmentation results by the physician,and the time consumption is lower.

关 键 词：测地线活动轮廓模型 灰度相似性信息 谱聚类算法 计算机断层扫描图像序列 肺实质分割 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]