AUTHOR=Zhao Feng , Chen Zhiyuan , Rekik Islem , Liu Peiqiang , Mao Ning , Lee Seong-Whan , Shen Dinggang 

TITLE=A Novel Unit-Based Personalized Fingerprint Feature Selection Strategy for Dynamic Functional Connectivity Networks

JOURNAL=Frontiers in Neuroscience

VOLUME=Volume 15 - 2021

YEAR=2021

URL=https://www.frontiersin.org/journals/neuroscience/articles/10.3389/fnins.2021.651574

DOI=10.3389/fnins.2021.651574

ISSN=1662-453X

ABSTRACT=The sliding-window-based dynamic functional connectivity networks (SW-D-FCN) constructed from resting-state functional Magnetic Resonance Imaging has become an increasingly useful tool in the diagnosis of various neurodegenerative diseases. However, what kind of strategy should be adopted to extract and select the most discrimination features from SW-D-FCN is still a challenging issue. Currently, selecting a better one from multiple feature sets and concatenating many kinds of feature sets are two of the most commonly used strategies. However, such strategies may fail to fully capture the personalized discriminative characteristics contained in each functional connectivity (FC) sequence of the SW-D-FCN. To address this issue, we propose a unit-based personalized fingerprint feature selection (UPFFS) strategy to better capture the most discriminative feature associated with disease for each unit. Specifically, the FC sequence between any pair of ROIs is regarded as a unit. For each unit, the most discriminative feature is selected from its multiple features by a specific feature evaluation method and all the most discriminative features are concatenated together as a feature set for subsequent classification. In such a way, the personalized fingerprint feature derived from each FC sequence can be fully mined and utilized in classification decision. To illustrate the effectiveness of the proposed strategy, we conduct experiments to identify the subjects with autism spectrum disorder from normal controls. Experimental results verify that the proposed strategy can select more discriminative features and achieve superior performance.