Extraction and Synchronization of BOLD Spontaneous Oscillations Using Singular Spectrum Analysis

Abstract

Spontaneous cerebral blood oxygenation level-dependent (BOLD) fluctuations are gaining interest in the neurophysiology community. These oscillations are prominent in the low-frequency range with spatiotemporal correlations. From a healthy individual, a basal resting state BOLD fMRI acquisition has been performed by collecting 4 slices. Voxel signals from seven selected regions have been considered. We assumed a composite null-hypothesis of oscillations embedded in “red noise”. To extract oscillations from BOLD signals we applied the Monte Carlo Singular Spectrum Analysis (SSA). Phase-synchronization of the oscillatory components, in the low-frequency range 0.085-0.13Hz, have been also achieved. As results, region-dependent distributions were apparent both for the noise parameters and for the number of connections between voxels. Although further studies on population samples should confirm the result consistency, the SSA technique combined with a phase-synchronization analysis seems a feasible method to extract low frequency BOLD spontaneous oscillations and to find functional connections among cerebral areas.