摘要
目的快速准确地分割脑部MR图像的海马结构对早期诊断阿尔兹海默氏症(Alzheimer's disease,AD)具有重要价值。该文提出了一种快速和准确分割MRI三维海马结构的方法。该方法基于格子玻尔兹曼(Lattice Boltzmann,LB)模型,利用脑部MR图像的边缘信息和区域信息建立一个三维分割模型(3D-LB),直接在三维空间中通过碰撞和迁移过程提取海马结构。为验证3D-LB分割模型的精度和效率,该文对30组海马结构的测试图像进行分割实验,并与三维CV模型进行比较。实验结果显示,基于3D-LB模型的分割方法能有效地分割海马结构的测试图像,且相较于三维CV模型,精度更高,所耗时间更少,表明LB方法适用于三维海马结构的快速和精确分割。
Fast and accurate segmentation of hippocampus in brain MRI is significantly helpful for radiolo- gists to diagnose Alzheimer＇s disease （AD） in early stage. This paper proposes a novel Lattice Boltzmann （LB） method to realize the segmentation. This method uses the edge and regional information of hippocampus, establishes a three - dimension LB model and realizes segmentation through the acting process of particles＂ collision and trans- form. Comparison experiments were carried out to test the accuracy and efficiency of the novel LB method. The no- vel LB method was compared to the CV method in 30 composite volume hippocampus MR images. The results showed that the novel LB method could achieve better performance and higher computation efficiency. Hence the feasibility of the novel LB method for segmentation of hippocampus images was proven.
出处
《生物医学工程学进展》
CAS
2014年第1期1-7,共7页
Progress in Biomedical Engineering
基金
国家自然科学基金(No.61171146)资助项目
