Multivariate empirical mode decomposition based sub-frequency bands analysis of the default mode network: a resting-state fMRI data study

Abstract

Abstract Resting-state functional connectivity reveals intrinsic, spontaneous networks that elucidate the functional architecture of the human brain. The default mode network (DMN) is the most important and stable intrinsic connectivity network (ICN), which involves several cognition functions, such as episodic memory and self-introspection. It has been suggested that low-frequency fluctuations in the blood oxygenation level-dependent (BOLD) signal during rest reflect the neuronal baseline activity of the brain and these low-frequency fluctuations correspond to functionally relevant resting-state networks. Several studies have revealed that the function of the brain is accomplished in certain low sub-frequency band. However, the concerned frequency bands are determined by experience, neglecting the intrinsic information of BOLD time series. In this study, we apply a full data-driven analysis, i.e., multivariate empirical mode decomposition (MEMD), to decompose resting-state fMRI data into the different sub-band DMNs, aiming to reveal the corresponding connectivity functions in separate sub-band DMN. Our results revealed that MEMD can adaptively decompose signals into intrinsic mode functions (IMFs) with the similar patterns across subjects. Furthermore, the sub-network constructed from the IMFs revealed that the different sub-band DMNs correspond to the different brain functional connectivity, inferring the possible relationships between sub-frequency band and cognitions. Owing to its data-driven merit, the proposed MEMD analysis may provide a new insight for fMRI-related studies.