融合感兴趣区域时空特征的阿尔茨海默症分类任务  

Classification Model for Alzheimer's Disease Based on Spatio-temporal Feature Fusion of ROIs

作　　者：李瑞 胡众义[1,2] 高礼彬 卢星进 

机构地区：[1]温州大学计算机与人工智能学院,浙江温州325035 [2]温州市智能影像处理与分析重点实验室,浙江温州325035

出　　处：《工业控制计算机》2023年第5期81-82,86,共3页Industrial Control Computer

基　　金：浙江省自然科学基金重大项目(LD21F020001);温州市科技计划重大科技创新攻关项目(ZY2019020);国家自然科学基金重点支持项目(U1809209)。

摘　　要：传统的功能连接网络模型只提取功能磁共振成像(functional Magnetic Resonance Imaging,fMRI)感兴趣区域(Regions Of Interest,ROIs)的时域特征,用于阿尔茨海默症(Alzheimer's Disease,AD)分类。该模型忽略了ROIs的空域特征,例如脑区之间的两两相关性。不完整的时空特征直接影响阿尔茨海默症、晚期认知障碍(late Mild Cognitive Impairment,lMCI)、早期认知障碍(early Mild Cognitive Impairment,eMCI)和健康对照(Healthy Controls,HC)分类任务的精确度。提出使用动态相关系数核(称为DC-kernel)对空域特征进行动态获取,并与时域特征进行特征融合。实验结果表明,在AD、lMCI、eMCI和HC四分类任务中,较传统阿尔茨海默症的准确率有提高,为其它脑疾病的分类任务提供了一种新的思路。The traditional functional connectivity network model extracts only the time-domain features of functional Magnetic Resonance Imaging(fMRI)Regions Of Interest(ROIs)for Alzheimer's Disease(AD)classification.The model ignores the null of ROIs.The model ignores the spatial features of ROIs,such as the two-by-two correlation between brain regions.Incomplete spatio-temporal features directly affect the accuracy of Alzheimer's disease,late Mild Cognitive Impairment(lMCI),early Mild Cognitive Impairment(eMCI)and Healthy Controls(HC)classification tasks.The accuracy of the classification tasks of lMCI,eMCI and HC is shown in this paper.This paper proposes to use a dynamic correlation coefficient kernel(called DC-kernel)for dynamic acquisition of spatial domain features and feature fusion with time-domain features.

关 键 词：功能连接网络 感兴趣区域 动态相关系数核 时空特征融合 阿尔茨海默症 分类任务 

分 类 号：R749.16[医药卫生—神经病学与精神病学] TP18[医药卫生—临床医学]