|本期目录/Table of Contents|

[1]魏红霞△,魏园园,王良敏,等.基于层次聚类的多模态磁共振脑肿瘤图像的自动分割方法*[J].生物医学工程研究,2021,02:144-148.
 WEI Hongxia,WEI Yuanyuan,WANG Liangmin,et al.Automatic segmentation method of multimodal magnetic resonance brain tumor images based on hierarchical clustering[J].Journal of Biomedical Engineering Research,2021,02:144-148.
点击复制

基于层次聚类的多模态磁共振脑肿瘤图像的自动分割方法*(PDF)

《生物医学工程研究》[ISSN:1006-6977/CN:61-1281/TN]

期数:
2021年02期
页码:
144-148
栏目:
出版日期:
2021-06-25

文章信息/Info

Title:
Automatic segmentation method of multimodal magnetic resonance brain tumor images based on hierarchical clustering
文章编号:
1672-6278 (2021)02-0144-05
作者:
魏红霞1△魏园园2王良敏3王力3许晓亮4
1.南阳市第二人民医院医学影像科,南阳 473000;2.南阳市第二人民医院神经内科,南阳 473000;3.南阳市第二人民医院放射诊断,南阳 473000;4.河南科技大学第一附属医院影像中心,洛阳 471000
Author(s):
WEI Hongxia1 WEI Yuanyuan2WANG Liangmin3WANG Li3XU Xiaoliang4
1.Department of Medical Imaging, Nanyang Second People′s Hospital,Nanyang 473000,China;2.Department of Neurology, Nanyang Second People′s Hospital, Nanyang 473000;3.Department of Diagnostic Radiology, Nanyang Second People′s Hospital, Nanyang 473000;4.Department of Imaging Center, the First Affiliated Hospital of Henan University of Science and Technology, Luoyang 471000,China
关键词:
层次聚类多模态磁共振脑肿瘤图像自动分割纹理特征
Keywords:
Hierarchical clustering Multimodal Magnetic resonance Brain tumor images Automatic segmentation Texture feature
分类号:
R318;TP394.41
DOI:
10.19529/j.cnki.1672-6278.2021.02.07
文献标识码:
A
摘要:
多模态磁共振脑肿瘤图像分割是医学图像应用的基础,在病理分析与手术引导等领域具有广泛的应用价值,为提升分割效率及精度,研究基于层次聚类的多模态磁共振脑肿瘤图像自动分割方法。利用Tamura的特征提取方法,以粗糙度与对比度为特征提取的定量分析指标,提取多模态磁共振脑肿瘤图像的纹理特征构建数据集;通过融合稀疏代表点的亲和传播聚类算法与密度峰值算法构建多层次聚类算法;利用稀疏代表点的亲和传播聚类算法粗分纹理特征数据集,获取代表点集;利用密度峰值算法聚类代表点集,合并粗分与聚类结果,完成肿瘤图像自动分割。实验证明,该方法能够精准分割多模态磁共振脑肿瘤图像,具有较高的自动分割效率。
Abstract:
Multimodal magnetic resonance brain tumor image segmentation is the basis of medical image application, and has a wide application in the fields of pathological analysis and surgical guidance. To improve the segmentation efficiency and accuracy, we studied an automatic segmentation method of multimodal magnetic resonance brain tumor image based on hierarchical clustering. Tamura′s feature extraction method, roughness and contrast were used as the quantitative analysis indexes of feature extraction, and the texture features of multimodal MRI brain tumor images were extracted to construct the data set.By the fusion of sparse representative point affinity propagation clustering algorithm and density peak algorithm,the multi-level clustering algorithm was constructed. The sparse representative points of affinity propagation clustering algorithm coarse texture feature dataset were used to obtain the representative point set. The peak density clustering algorithm was used to cluster the representative points, the rough and clustering results were combined to complete the automatic segmentation of tumor image. Experimental results show that this method can accurately segment multimodal MRI brain tumor images and has high automatic segmentation efficiency.

参考文献/References

备注/Memo

备注/Memo:
(收稿日期:2020-11-30)河南省医药卫生基金资助项目(2020A258)。△通信作者Emial:wllx8505@163.com
更新日期/Last Update: 2021-07-21