Functional Connectivity of Cortical Resting-State Networks to the Thalamus

Introduction: The thalamus is made up of a number of different nuclei, which are closely connected to each other by association fibers and to cortical and subcortical brain regions by projection fibers. Almost all sensory and somatosensory pathways except olfactory afferences project to the contralateral thalamus and are then forwarded to the cerebral cortex. Therefore, the thalamus can be seen as a central relay and integration center of the CNS, which serves as the gateway to the telencephalon (Jones, 2007; Sherman and Guillery, 2009). In the resting state fMRI (rfMRI) a variable number of cortical networks (c-RSN) have been described. The ten most prominent c-RSNs were reported in detail by Smith et al. (Smith et al., 2009). As the thalamus temporally binds to various cortical areas for sensory computations and cortico-cortical-feedback, it is of interest to examine how these cortical networks communicate with thalamus during rest. We hypothesized that all c-RSNs differently communicate to various thalamic nuclei during rest and that these functional connections may be organized in a distinct spatial pattern. In the presented work, we examined the Smith-10 cortical c-RSN on their functional connectivity with the thalamus in a large data set of 730 subjects of the Human Connectome Project (HCP) using a "winner takes it all" (WTA) approach. Methods: Four resting state sessions (gradient-echo EPI, 1200 scans/session, duration: 14:33 min, TR: 720 ms, TE: 33.1 ms, resolution: 2 mm isotropic, 72 slices, multiband factor: 8 of 730 subjects were chosen from the HCP dataset (Van Essen et al., 2012). Data were preprocessed and ICA denoised using the available HCP pipeline (Glasser et al., 2013) and FSL Fix. Data were smoothed with 3.5 mm kernel. Ten c-RSN templates from Smith et al. (Smith et al., 2009) were used for classification of thalamic voxels by performing correlation analysis using the fsl (O'Reilly 2009). The correlation group maps were calculated using a fixed-effect analysis. The fixed effect group maps were then used to calculate the dominant thalamic representation. Results: The winner-takes-it-all (WTA) maps of the ten c-RSNs showed a topographic organization within left and right thalamus (s. Fig. 1), in which the sensorimotor and cerebellum networks displayed a higher degree of spatial spread compared to the other networks. A detailed comparison revealed that five thalamic RSNs clusters exhibited noticeable hemispheric differences for the sensorimotor, cerebellum, and fronto-parietal-left network to the left and the executive and right fronto-parietal network to the right thalamus