Effects of Partial Sleep Deprivation on the Resting-State Dynamic Functional Connectivity

Sleep deprivation (SD) affects many cognitive and emotional functions [1]. While conventional resting state fMRI studies have shown changes in the overall functional connectome of the brain following SD [1], the dynamic characteristics of these changes have barely been examined. Dynamic functional connectivity (DFC) has recently been proposed to study the instantaneous co-activation of different brain regions and many mathematical methods have been developed to assess it including phase coherence [2]. Method We used the resting state fMRI data of 61 healthy subjects. The data included two sessions that were recorded with an interval of one month; in one the subjects underwent partial SD (3h sleep) and in the other they had full sleep [3]. Data was preprocessed with DPARSF. We computed the instantaneous phase from the Hilbert transform of the BOLD signal for each region at each TR and then calculated dFC using Eq1. (ni is a brain region, t is a TR, φ is the phase of the signal) [2]. This leads to an N×N dFC matrix for each TR (N = number of regions). The leading Eigenvectors of each matrix was extracted, k-means was used to cluster the vectors and FC states were reconstructed using the cluster centroids. Probability of occurrence and lifetime of the states were calculated and a permutation-based paired t-test was applied to test for between-group significance.Eq1. ├ dFC(n_1.n_2.t)=cos⁡(φ(n_1.t)┤ -⁡φ (n_2.t))Results The results showed significant difference in the lifetime of three FC states and the probability of occurrence of one FC state between the SD and full sleep conditions. These three states, their probability and lifetime for the two conditions, and the p values are shown in Figure 1.Conclusions SD affects the temporal behavior of FC patterns in the brain, including how often a certain FC state occurs and how long it lasts before another state dominates the patterns.