Feasibility and reproducibility of resting-state functional magnetic resonance imaging in asphyxiated neonates.

BACKGROUND

In the ongoing search for prognostic tools in neonates suffering perinatal asphyxia, we assessed the feasibility of resting-state functional magnetic resonance imaging (RS-fMRI). The aim of our study was to determine both the intra-session and inter-subject reproducibility of resting-state networks (RSNs), as well as factors affecting these.

NEW METHOD: We conducted a prospective multicenter study in 21 asphyxiated term newborns (mean gestational age = 39.2 weeks, mean 5-minute Apgar score = 3) after treatment with controlled hypothermia at a level III neonatal intensive care unit. RSNs were assessed twice with RS-fMRI at 1.5 T using both group MELODIC ICA and single-subject ICA. Intraclass correlation coefficients were calculated on all networks identified from the group MELODIC. Feasibility of the RS-fMRI technique was determined by assessing the ability to discriminate common RSNs, the presence of motion artifacts and the temporal signal to noise ratio.

RESULTS

Seven RSNs could be identified, anterior default mode network aDMN), default mode network (DMN), left frontoparietal network (lFPN), right frontoparietal network (rFPN), auditory network (AN), sensorimotor network (SMN), parietal network (ParietalN). Two networks (DMN and rFPN) demonstrated moderate reproducibility (ICC 0.62 and 0.69, respectively). The other networks demonstrated poor reproducibility (ICC <0.5).

CONCLUSION

Following perinatal asphyxia, reproducibility was moderate for DMN and rFPN and poor for the other RSNs.