Functional MRI is increasingly being used to study the emerging functional architecture of the brain in the neonatal period. Previous resting-state fMRI studies reported that, although the majority of networks are present at the time of normal birth, they appear to mature at different rates with the "higher-order" networks (such as default mode network) developing later than primary sensory networks (such as sensori-motor, auditory, and visual networks)¹.This trend is also seen in structural MRI studies during early development, which have shown that sulcal maturation progresses initially around the central sulci, followed by temporo-parieto-occipital regions, with the frontal lobe developing last².T₂* values in neonates are up to 2 times longer than those typically seen in the mature adult brain³ and decrease with increasing gestational age at scan During the preterm period (equivalent to the third trimester), there are marked increases in synaptic density and cerebral vasculature and a general reduction in brain water content, all of which may contribute to the maturational decrease in T₂* values seen at this time^3,4. In this study we assessed T_²* values for different resting-state networks (RSNs) and hypothesized that those which mature earlier on structural MRI would have shorter T₂* values than networks in cortical regions that develop later in gestation.The study group consisted of 7 healthy term-born infants (median post-menstrual age 40.65 weeks). Written parental consent was obtained for all subjects. Data was acquired on a Philips 3T Achieva system (Best, NL) with a 32 channel receiver head coil, using a feed-and-wrap technique without sedation. In each infant, 3 dual-echo resting state echo BOLD fMRI data sets were acquired using an Echo-planar-imaging (EPI) sequence (TR = 2.998 s, flip angle 76o, in-plane resolution 2.5x2.5mm, slice thickness 3mm, 5 min. total acquisition time, TE1/TE2= 25/85,45/95,65/114ms). The data were analysed using tools as implemented in FMRIB's software library (www.fmrib.ox.ac.uk/fsl). Data were pre-processed using a standard pipeline which included removal of non-brain tissue, rigid-body motion correction, spatial smoothing (FWHM 3mm), and affine registration to a neonatal standard space template4. For each individual subject, ICA decomposition was first performed using MELODIC⁵, and components identified as those related to noise artifact were regressed out (denoising). Manual identification of noise components was performed by a rater (MA), blinded to the subject/echo time, and classified them as related to physiological and motion artifacts, head motion, or image distortion on the basis of their spatial and temporal characteristics. Further group level analysis was then performed on the denoised data-sets using MELODIC by temporally concatenating the individual, co- registered subject data (across all the TEs). T₂* maps were obtained by monoexponential fitting (Mtlab 2013b) to the average of the first 25 resting-state volumes, from 6 echo times, for each subject. T₂-weighted images were segmented into grey matter (cortex and deep grey matter), white matter and CSF using GUITEST software⁶ (Figure 1). The segmented grey matter and group RSN maps were registered to T₂* images and were used as masks (to exclude WM and CSF) to calculate mean T₂* for each RSN per subject. ANOVA was performed to assess significant variations between RSN T₂* values(p<0.05).One data set was discarded due to excessive motion during acquisition. Group melodic identified 9 resting state including Motor cortex, Auditory, Thalamus, Basal Ganglia, Parietal Frontal Cortex and Default Mode Networks. ANOVA analysis showed that there is a significant variation in T₂* across different RSNs (Figure 2).RSNs could be reliably identified across our whole study group. The absolute value of the T₂* strongly affects fMRI sensitivity and this factor is of particular importance for carrying out an optimized neonatal fMRI experiment, as blood oxygen level dependent contrast-to-noise ratio is known to be maximal when the acquisition sequence echo time is matched to the T₂* of the tissue of interest⁷. Our results suggest that, in healthy term neonates, there is a regional network-specific T₂* variation. Our finding is in agreement with a previous structural MRI study⁸ that showed the development of the central sulcus and medial occipital lobe (equivalent to visual and motor networks) was more advanced at term , while an intermediate level of gyral development was found in the posterial temporal and parietal lobes; the areas where auditory and parietal temporal cortex are located. Frontal and anterior temporal lobes (equivalent to default mode) were the last to develop. The results also suggest that these regional differences in T₂* should be considered in future neonatal fMRI studies, and echo-time could be adapted to optimize sensitivity for particular brain networks depending on the specific research question.