南京理工大学学报（自然科学版）2019，Vol.43Issue(5)：578-585,8.DOI:10.14177/j.cnki.32-1397n.2019.43.05.006
基于多核协同表示分类的脑肿瘤分割算法
Brain tumor segmentation algorithm based on multi-kernel collaborative representation classification
葛婷 1詹天明 2牟善祥3
作者信息
- 1. 南京理工大学 电子工程与光电技术学院,江苏 南京210094
- 2. 金陵科技学院 理学院,江苏 南京211169
- 3. 南京审计大学 信息与工程学院,江苏 南京211815
- 折叠
摘要
Abstract
In order to segment brain tumor regions from brain magnetic resonance( MR) image and provide reference for subsequent disease diagnosis and surgical navigation, a brain tumor segmentation algorithm is proposed based on the multi-kernel collaborative representation classification under the framework of kernel method. Firstly,multi-scale superpixel segmentations of brain tumor images are carried out and the spatial features based on superpixel regions are constructed. Then the original spectral information and the extracted multi-scale spatial features are fused by using the multi-kernel collaborative representation classification method under the multiple kernel frame work. Finally,the segmentation of brain tumor regions is realized in combination with clinical features. Test results on the data sets of MICCAI BraTS 2012 and 2013 show that,compared with the existing brain tumor segmentation algorithms,the proposed method can extract brain tumor regions better and has better segmentation accuracy.关键词
核磁共振图像/脑肿瘤/图像分割/超像素/多尺度/多核协同表示分类Key words
magnetic resonance images/brain tumors/image segmentation/superpixel/multi-scales/multi-kernel collaborative representation classification分类
信息技术与安全科学引用本文复制引用
葛婷,詹天明,牟善祥..基于多核协同表示分类的脑肿瘤分割算法[J].南京理工大学学报（自然科学版）,2019,43(5):578-585,8.
