基于张量子空间的半脑对称度特征与癫痫识别

发布时间：2018-03-05 17:51

本文选题：癫痫　切入点：张量　出处：《东北大学学报(自然科学版)》2017年07期 　论文类型：期刊论文

【摘要】：结合脑PET图像信息,提出了一种基于张量子空间的半脑对称度特征的识别方法用于识别PET图像中癫痫病灶.首先计算全部脑PET图像中所有体素的SUV,并基于SUV建立三阶张量;然后提取半脑对称度特征,建立半脑对称度张量模型;其次利用多线性主成分分析(MPCA)方法对半脑对称度张量模型进行特征选择;最后基于支持向量机(SVM)分类器进行癫痫识别.实验结果表明:提出的算法能够有效地识别脑PET图像中的癫痫病灶,可以作为计算机辅助诊断方式帮助医生进行癫痫疾病的诊断.
[Abstract]:Based on the information of brain PET images, a method of recognition of hemispheres symmetry based on tensor quantum space is proposed to identify epileptic foci in PET images. Firstly, the sum of all voxels in all brain PET images is calculated, and the third-order Zhang Liang is established based on SUV. Then the hemispheres symmetry degree feature was extracted, and the Zhang Liang model of hemisencephalic symmetry degree was established. Secondly, the feature selection of the hemisencephalic symmetry degree Zhang Liang model was carried out by using the multi-linear principal component analysis (MPCA) method. The experimental results show that the proposed algorithm can effectively identify epileptic foci in brain PET images and can be used as a computer-aided diagnosis method to help doctors diagnose epilepsy.
【作者单位】：东北大学软件学院;东北大学中荷生物医学与信息工程学院;
【基金】：国家自然科学基金资助项目(61472073)
【分类号】：R742.1;TP391.41

【相似文献】