基于改进窄带水平集的三维肝脏肿瘤分割

三维肝脏肿瘤识别是当前研究的热点问题，如何准确快速地从腹部CT序列中分割出肝脏肿瘤是肝部病变诊断的基础。针对水平集方法在进行分割时收敛速度较慢，设置窄带宽度固定不灵活的缺点，先利用分水岭算法，对肝脏图像进行“过分割”，搜索初始轮廓所在的分水岭块作为窄带区域进行标记，在窄带区域内用水平集算法使初始轮廓线收敛至准确轮廓。再以其边缘作为相邻CT序列的肿瘤初始轮廓，找出初始轮廓线所在的分水岭块，构成新的窄带，用水平集算法对轮廓线进行迭代分割出肿瘤。重复该过程，直至完成整个肝脏序列图像的肿瘤图像分割，进行三维重建。

3D segmentation of liver tumors combined with improved narrow band level set

Three-dimensional liver tumor segmentation is a hot issue in medical image processing study. Accurate and fast liver tumor segmentation from abdominal CT sequences is the basis for liver lesions diagnose. The level set method is sensitive to contour and has large amount of computation. It is inflexibility to set the width of the narrow band. The new method improves traditional level set method. Firstly, divided liver image with watershed method, the block where initial contour in is marked, and the labeled blocks make up narrow band. The level set algorithm can convergence initial contour to the exact contour in narrow band. Then using its edge as the initial contour of the adjacent CT sequence, with the new algorithm mentioned above split the liver tumor from a sequence of images, the process is repeated until get all segmentation result from the entire slices, then 3-D reconstructed.