CAJ | 학술논문

由于人体腹部CT图像纹理结构复杂，灰度差别特别小，边缘又很不明显，本文提出了基于区域生长法的序列图像分割算法。在提取肝脏图像时，将全部序列图像分成若干组，然后利用区域生长算法先分割出每组的第一张图片。对于本组内其余图片的分割，充分利用相邻图片的灰度相似性实现序列图像的分割。对于血管的分割，由于血管在三维空间上的连通性，采用三维区域生长算法实现血管管道的提取。实验结果表明，应用本文算法能得到准确的肝脏及血管分割结果。
유우인체복부CT도상문리결구복잡，회도차별특별소，변연우흔불명현，본문제출료기우구역생장법적서렬도상분할산법。재제취간장도상시，장전부서렬도상분성약간조，연후이용구역생장산법선분할출매조적제일장도편。대우본조내기여도편적분할，충분이용상린도편적회도상사성실현서렬도상적분할。대우혈관적분할，유우혈관재삼유공간상적련통성，채용삼유구역생장산법실현혈관관도적제취。실험결과표명，응용본문산법능득도준학적간장급혈관분할결과。
For the complex characteristics of medical image , a segmentation algorithm based on region growing algorithm is proposed in this paper , which is applied to segment a series of CT images .This method is implemented to segment liver CT slices .Firstly, all slices are divided into groups;the first slice of every group is segmented .Fi-nally, the other adjacent slices in every group are segmented on the basis of gray shade similarity .According to 3D connectivity of blood vessels , a 3 D region growing algorithm is applied to segment blood vessels .As the experiment results show , the segmentation results of liver image are accurate by using adaptive region growing , and this algo-rithm can provide accurate data for 3 D reconstruction and surgical simulation of liver and blood vessels .