太赫兹科学与电子信息学报, 2020, 18 (4): 698, 网络出版: 2020-12-25  

基于脑网络组图谱的阿尔茨海默病分类

Classification analysis for Alzheimer’s Disease based on human brainnetome atlas
作者单位
1 哈尔滨理工大学自动化系, 黑龙江哈尔滨 150080
2 中国科学院自动化研究所模式识别国家重点实验室, 北京 100190
3 山东师范大学信息科学与工程学院, 山东济南 250014
4 中国人民解放军总医院南楼神经内科, 北京 100853
摘要
结构磁共振影像 (sMRI)作为一种非入侵式的脑成像方式为人们理解阿尔茨海默病 (AD)的患病机制提供了很大的帮助, 目前已有大量研究利用从 sMRI中提取的特征进行 AD的识别。为了充分利用图像信息提取 AD相关的特征, 提出了一种简单易用的基于人类脑网络组图谱的脑区划分进行特征提取的方法。选取美国阿尔茨海默病神经影像组织 (ADNI)数据库中的 226例正常被试 (NC)和227例 AD患者的sMRI数据作为研究对象, 提取每个个体脑网络组图谱中各脑区的平均灰质密度作为特征, 利用支持向量机 (SVM)对NC和AD患者进行分类, 通过 10折交叉验证的方式得到了 85.2%的平均分类准确率。后续的统计分析发现, 海马、杏仁核及梭状回等脑区的平均灰质密度对 NC和AD患者的识别贡献很大, 且这些脑区的萎缩程度与患者的认知能力密切相关。利用最小绝对收缩选择算子 (LASSO)对个体的简易智力状态评分 (MMSE)进行预测, 预测结果与真实值间存在显著的正相关 (r0.65, p0.001)。研究结果表明, 基于脑网络组图谱脑区划分提取的脑区平均灰质密度特征可以有效地对 AD患者进行识别, 并可以用来评估个体的认知水平。
Abstract
As a noninvasive brain imaging method, structural Magnetic Resonance Imaging(sMRI) plays an important role in understanding the pathology of Alzheimer’s Disease(AD). Convergence evidence suggests that imaging features based on sMRI performs well in classifying AD patients from Normal Controls(NC). In this paper, a simple feature extraction method is proposed based on the Human Brainnetome Atlas with 227 AD patients and 226 NC from Alzheimer’s Disease Neuroimaging Initiative (ADNI) database. Firstly, mean gray matter density of each brain region in the brainnetome atlas is obtained as feature. After that, an Support Vector Machine(SVM) model is introduced to classify AD from NC samples. The result shows that the mean accuracy is 85.2% with 10-fold cross validation. And post hoc analysis demonstrates that the mean gray matter density of several brain regions such as the hippocampus, amygdala and fusiform gyrus play important roles in classification and the atrophy of these regions has significant correlation with cognitive ability measured by the Mini-Mental State Examination(MMSE) scores in AD patients. In addition, the Least Absolute Shrinkage and Selection Operator(LASSO) is utilized to predict individual MMSE score and the predicted values show a very high positive correlation with true values (r0.65, p0.001). In conclusion, the present study demonstrates that the gray matter density of brain regions defined by the Brainnetome Atlas is meaningful for distinguishing AD patients from NC samples and could also be utilized to predict individual cognitive ability.

胡方舟, 尤波, 赵坤, 李卓然, 丁艳辉, 张熙, 刘勇. 基于脑网络组图谱的阿尔茨海默病分类[J]. 太赫兹科学与电子信息学报, 2020, 18(4): 698. HU Fangzhou1, YOU Bo, ZHAO Kun, LI Zhuoran, DING Yanhui, ZHANG Xi, LIU Yong. Classification analysis for Alzheimer’s Disease based on human brainnetome atlas[J]. Journal of terahertz science and electronic information technology, 2020, 18(4): 698.

关于本站 Cookie 的使用提示

中国光学期刊网使用基于 cookie 的技术来更好地为您提供各项服务,点击此处了解我们的隐私策略。 如您需继续使用本网站,请您授权我们使用本地 cookie 来保存部分信息。
全站搜索
您最值得信赖的光电行业旗舰网络服务平台!