自适应形状约束Graph cuts算法在腹部CT图像分割中的应用
|
摘要
针对CT图像对比度低,组织边缘模糊,器官轮廓不规则等特点造成组织器官难以分割的问题,提出了一种自适应形状约束的Graph cuts算法.首先使用基于多图谱配准的分割方法分割原始图像,将得到的初始分割结果作为形状先验加入到Graph cuts算法的能量函数中,同时根据配准分割过程得到的目标概率图自适应选择形状约束项系数,最后通过最大流最小割算法分割出CT图像中的肝脏、肾脏和脾脏.实验结果表明,该方法能够较好地分割出肝脏等组织器官,有效减轻传统图割算法分割图像时造成的过分割和欠分割现象.
Because of the low contrast, the blurred edges and the irregular contours of the organs in CT images, it is hard to extraction the target organ from CT images. To address this problem, a graph cuts algorithm which combines shape constraint adaptively is proposed. Firstly, the original image is segmented using the segmentation method based on multi-atlas registration, and the initial segmentation is obtained. Then the initial segmentation is added to the graph cuts energy function as a shape constraint. At the same time, the shape constraint coefficients are adaptively selected according to the target probability map which obtained in the process of initial segmentation. Finally the targets such as liver, kidney and spleen in the CT image are extracted by the max-flow min-cut algorithm which minimize the energy function. The experimental results show that this method can segment the targets well, and it can effectively reduce the over-segmentation and under-segmentation caused by the traditional graph cut algorithms.
Because of the low contrast, the blurred edges and the irregular contours of the organs in CT images, it is hard to extraction the target organ from CT images. To address this problem, a graph cuts algorithm which combines shape constraint adaptively is proposed. Firstly, the original image is segmented using the segmentation method based on multi-atlas registration, and the initial segmentation is obtained. Then the initial segmentation is added to the graph cuts energy function as a shape constraint. At the same time, the shape constraint coefficients are adaptively selected according to the target probability map which obtained in the process of initial segmentation. Finally the targets such as liver, kidney and spleen in the CT image are extracted by the max-flow min-cut algorithm which minimize the energy function. The experimental results show that this method can segment the targets well, and it can effectively reduce the over-segmentation and under-segmentation caused by the traditional graph cut algorithms.
引文
[1] 邢东炜.螺旋CT测量肝硬化后肝脾肾体积变化的研究进展[M].全国第十三次中西医结合影像学术研讨会暨福建省第八次中西医结合影像学术研讨会论文集.武夷山:中国中西医结合学会,2014.
[2] 赵于前,王小芳,李桂源.基于多尺度多结构元素的肝脏图像分割[J].光电子·激光, 2009(4):563-566.
[3] 何兰,吴倩.基于3D卷积神经网络的肝脏自动分割方法[J].中国医学物理学杂志,2018(6):680-686.
[4] 潘晓佩.基于融合算法的超声肾脏图像分割[J].现代计算机(专业版),2016(3):27-32.
[5] CHEN X, UDUPA J K, BAGCI U, et al. Medical image segmentation by combining graph cut and oriented active appearance models[J]. IEEE Transactions on Image Processing A Publication of the IEEE Signal Processing Society, 2012, 21(4):2035-2046.
[6] BOYKOV Y,KOLMOGOROV V. An experimental com-parison of min-cut/max-flow algorithms for energy minimization in vision.[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2004, 26(9):1124-1137.
[7] BOYKOV Y Y, JOLLY M P. Interactive graph cuts for optimal boundary & region segmentation of objects in N-D images[C]// IEEE.IEEE International Conference on Computer Vision. Vancouver: IEEE Computer Society, 2001:105-112.
[8] 周亭亭.基于多图谱分割的融合算法研究[D]. 银川:宁夏大学,2017.
[9] JR C R M, QI R, RAGHAVAN V. A linear time algorithm for computing exact euclidean distance transforms of binary images in arbitrary dimensions[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2003, 25(2):265-270.
[10] FREEDMAN D, ZHANG T. Interactive graph cut based segmentation with shape priors[C]//IEEE. IEEE Computer Society Conference on Computer Vision and Pattern Recognition. San Diego: IEEE Computer Society, 2005:755-762.
[11] KLEIN S,VAN-DER-HEIDE U,LIPS I,et al.Automatic segmentation of the prostate in 3D MR images by atlas matching using localized mutual information[J]. Medical Physics, 2008, 35(4):1407-1417.
[12] KLEIN A,ANERSSON J,ARDEKANI B A,et al. Evalua-tion of 14 nonlinear deformation algorithms applied to human brain MRI registration.[J]. Neuroimage, 2009, 46(3):786-802.
[13] SETHIAN J A. Level set methods and fast marching method[M]. Cambridge: Cambridge University Press, 1999.
[2] 赵于前,王小芳,李桂源.基于多尺度多结构元素的肝脏图像分割[J].光电子·激光, 2009(4):563-566.
[3] 何兰,吴倩.基于3D卷积神经网络的肝脏自动分割方法[J].中国医学物理学杂志,2018(6):680-686.
[4] 潘晓佩.基于融合算法的超声肾脏图像分割[J].现代计算机(专业版),2016(3):27-32.
[5] CHEN X, UDUPA J K, BAGCI U, et al. Medical image segmentation by combining graph cut and oriented active appearance models[J]. IEEE Transactions on Image Processing A Publication of the IEEE Signal Processing Society, 2012, 21(4):2035-2046.
[6] BOYKOV Y,KOLMOGOROV V. An experimental com-parison of min-cut/max-flow algorithms for energy minimization in vision.[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2004, 26(9):1124-1137.
[7] BOYKOV Y Y, JOLLY M P. Interactive graph cuts for optimal boundary & region segmentation of objects in N-D images[C]// IEEE.IEEE International Conference on Computer Vision. Vancouver: IEEE Computer Society, 2001:105-112.
[8] 周亭亭.基于多图谱分割的融合算法研究[D]. 银川:宁夏大学,2017.
[9] JR C R M, QI R, RAGHAVAN V. A linear time algorithm for computing exact euclidean distance transforms of binary images in arbitrary dimensions[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2003, 25(2):265-270.
[10] FREEDMAN D, ZHANG T. Interactive graph cut based segmentation with shape priors[C]//IEEE. IEEE Computer Society Conference on Computer Vision and Pattern Recognition. San Diego: IEEE Computer Society, 2005:755-762.
[11] KLEIN S,VAN-DER-HEIDE U,LIPS I,et al.Automatic segmentation of the prostate in 3D MR images by atlas matching using localized mutual information[J]. Medical Physics, 2008, 35(4):1407-1417.
[12] KLEIN A,ANERSSON J,ARDEKANI B A,et al. Evalua-tion of 14 nonlinear deformation algorithms applied to human brain MRI registration.[J]. Neuroimage, 2009, 46(3):786-802.
[13] SETHIAN J A. Level set methods and fast marching method[M]. Cambridge: Cambridge University Press, 1999.