论文部分内容阅读
针对现有二维分割方法人工干预较多及存在分割缺陷、三维分割方法对突变异常肝脏分割错误等问题,本文提出一种基于图像序列上下文关联的肝脏器官半自动分割方法。利用肝脏器官组织图像序列上下文的相似性先验知识,结合区域生长和水平集模型,并以少量人工干预辅助应对肝脏突变情况来进行肝脏的半自动分割。实验结果表明,本文方法分割肝脏精度高,适应能力强,对变异性较大的肝脏分割效果较好,可较好地满足临床应用需求。
Aiming at the problems of the existing two-dimensional segmentation methods, such as more manual intervention and segmentation defects, and the three-dimensional segmentation method to the mutation abnormal liver segmentation errors, this paper proposes a semi-automatic segmentation method of liver organ based on image sequence context association. Using the priori knowledge of the similarity of the image sequence of liver organ tissues, combining the model of regional growth and level set, the liver was semi-automatically segmented by a small amount of manual intervention to assist in the treatment of liver mutation. The experimental results show that the proposed method has the advantages of high accuracy and adaptability in the liver, good effect on liver segmentation with large variability, which can meet the needs of clinical application.