文章基本信息

标题：Multiple Connection Pattern Combination for Mild Cognitive Impairment Identification from Single Modal Data
本地全文：下载
作者：Wei-Kai Li ; Xiao-Wen Xu ; Xin Gao 等
期刊名称：IFAC PapersOnLine
印刷版ISSN：2405-8963
出版年度：2020
卷号：53
期号：5
页码：7-12
DOI：10.1016/j.ifacol.2021.04.077
语种：English
出版社：Elsevier
摘要：AbstractAccurate diagnosis of Mild Cognitive Impairment (MCI) has gained much more attention in the past few years. It should be noted that, in many recent works, the brain connection estimated by fMRI data have been provided the effective and robust biomarker for several neurological-disorders diagnosis. However, the existing works only focus on the single connection pattern (e.g., Pearson’s Correlation) for diagnosis, which often ignores the information from other patterns for diagnosis. Consequently, such approaches may not be sufficient to reveal the underlying connection differences between the groups of disease-affected patients and normal controls (NC), which limited its performance. As a result, Multiple Connection Pattern Combination (MCPC) method from Single Modal Data is come into beings. Specifically, we propose to combine three patterns of connections including Pearson’s Correlation, Partial Correlation and Dynamic Casual Modelling to identify MCI from Normal Controls (NC), using a kernel combination trick. 68 MCI and 69 NC from ADNI dataset are used for development and validation of our proposed MCPC method. As a result, the proposed methods achieved 87.40 % classification accuracy, significantly outperformed the case of methods using the single connection pattern.
关键词：KeywordsFunctional ConnectivityMulti-ViewMulti-ModalMild Cognitive Impairment