Examining Brain Morphometry Associated with Self-Esteem in Young Adults Using Multilevel-ROI-Features-Based Classification Method.
Examining Brain Morphometry Associated with Self-Esteem in Young Adults Using Multilevel-ROI-Features-Based Classification Method.
复制标题
DOI：
10.3389/fncom.2017.00037
复制
发表时间：
2017
影响因子：
3.2
通讯作者：
Dai Y
中科院分区：
文献类型：
--
作者：
Peng B;Lu J;Saxena A;Zhou Z;Zhang T;Wang S;Dai Y
目的：采用基于多水平特征的脑磁共振图像分类方法，研究自尊相关的脑形态特征。方法：多层次感兴趣区域（ROI）特征由两类特征组成：(i)感兴趣区域特征，包括灰质体积、白质体积、脑脊液体积、皮质厚度和皮质表面积；（ii）相似特征，基于ROI之间皮质厚度的相似度计算。针对每种特征类型，采用基于滤波器和基于包装的混合特征选择方法来选择最具判别性的特征。利用多核支持向量机（svm）对ROI特征和相似特征进行融合，并赋予相应的权重因子。结果：多级ROI特征提高了分类性能，准确率为96.66%，特异性为96.62%，灵敏度为95.67%。与自尊相关的最具区别性的ROI特征分布在枕叶、额叶、顶叶、边缘叶、颞叶和中央区域，主要涉及白质和皮层厚度。最具区别性的相似性特征分布在左右半球，包括额叶、枕叶、边缘叶、顶叶和中央区域，传递了大脑不同区域之间的结构连接信息。结论：本文利用ROI特征和相似特征检测自尊相关的脑形态测量，为自尊与特定ROI和不同脑区之间的结构连接相关提供了初步证据。
Purpose: This study is to exam self-esteem related brain morphometry on brain magnetic resonance (MR) images using multilevel-features-based classification method. Method: The multilevel region of interest (ROI) features consist of two types of features: (i) ROI features, which include gray matter volume, white matter volume, cerebrospinal fluid volume, cortical thickness, and cortical surface area, and (ii) similarity features, which are based on similarity calculation of cortical thickness between ROIs. For each feature type, a hybrid feature selection method, comprising of filter-based and wrapper-based algorithms, is used to select the most discriminating features. ROI features and similarity features are integrated by using multi-kernel support vector machines (SVMs) with appropriate weighting factor. Results: The classification performance is improved by using multilevel ROI features with an accuracy of 96.66%, a specificity of 96.62%, and a sensitivity of 95.67%. The most discriminating ROI features that are related to self-esteem spread over occipital lobe, frontal lobe, parietal lobe, limbic lobe, temporal lobe, and central region, mainly involving white matter and cortical thickness. The most discriminating similarity features are distributed in both the right and left hemisphere, including frontal lobe, occipital lobe, limbic lobe, parietal lobe, and central region, which conveys information of structural connections between different brain regions. Conclusion: By using ROI features and similarity features to exam self-esteem related brain morphometry, this paper provides a pilot evidence that self-esteem is linked to specific ROIs and structural connections between different brain regions.