基于GPU的膝盖组织分割与显示方法-《中国医学物理学杂志》

文章信息/Info

Author(s):: JIANG Xiangang1; ZHANG Hongbin2; 1. School of Science, East China Jiaotong University, Nanchang 330013, China 2. School of Software, East China Jiaotong University, Nanchang 330013, China

Keywords:: Keywords: knee arthritis meniscus 3D segmentation diffusion-based region growing algorithm

摘要:: 针对膝盖器官组织在多模态图像中存在复杂的粘连和交错叠加的情况，提出在扩散区域增长算法中增加种子亮度限制参数并与边缘限制参数一起共同控制区域增长的迭代过程，该算法和相关的滤波算法通过采用GPU编程方式实现，并将半月板临近组织的分割从股骨和胫骨的最近连接处展开。实验结果表明，本方法对于膝盖组织和相关病灶的分割提供可视的实时显示，对膝盖组织相关疾病的微创手术和介入治疗提供导视。【关键词】膝盖；关节炎；半月板；3D分割；扩散区域增长算法

Abstract:: Abstract: Considering that knee tissues have complex adhesion and overlapping in multimodal image, a new diffusion-based region growing algorithm combining edge constrain and newly-added seed point luminance constrain is presented to control the iteration of region growing. The algorithm and relevant filtering algorithms are implemented by GPU. The proposed method focuses on searching the nearest connection between the femur and tibia for meniscus and parenchyma segmentations. The experiment results show that the proposed method can provide virtual reality for the segmentations of knee tissues and relevant lesions, and offer guidance for minimally invasive surgery and interventional therapy for knee diseases.