| dc.description.abstract | In recent years, the improved survival of very/extremely preterm infants has led to a higher prevalence of neurodevelopmental impairment. As such, increased attention has turned towards biomarkers for early diagnosis/prognosis. Historically, neurological insults during the early preterm period (while in the neonatal intensive care unit) took the form of macroscopic lesions leading to cerebral palsy, blindness/deafness, or other severe motor/cognitive impairment. However, today, very/extremely preterm infants mostly avoid macroscopic brain injury and are instead later diagnosed with more subtle deficits, such as disorders of attention, executive function, or learning. Here we investigate how functional neuroimaging can be used to probe individual differences in infant brain function. Specifically, we examine functional connectivity. In the first study, we critically examine existing methods for measuring the individual functional connectome (a.k.a. `fingerprint'), which is portion of functional connectivity that can identify the individual and be linked to their behaviours. Recent research has assessed whether a functional fingerprint exists in young infants. We hypothesised that various confounding variables, including brain state (asleep versus awake), might impact this functional connectome in infants (and its potential as a neuroimaging biomarker). Analysing data from a large open infant cohort, we identified confounding variables that seriously impact infant functional fingerprinting, including motion, head size, inter-session interval and head position in the head coil. Next, we acquired functional MRI (fMRI) from a cohort of infants as part of the FOUNDCOG study (www.foundcog.org). The first 100 infants (75 term and 25 preterm born infants) were included in this analysis. The majority of previous studies have scanned infants asleep, and the practical challenges and important methodological steps needed to achieve awake infant fMRI data is discussed in detail. Methods used resulted in successful completion of awake fMRI runs and good infant comfort/attendance (as reported by researcher and parents). Individual versus shared components of the infant functional connectome were identified in preterm and term infant groups, and the impact of brain state (awake versus asleep) on these was assessed. Results from this 2-month old infant cohort, showed that there is a different pattern of fMRI connectivity between sleep and awake (movies) states, and that there appear subtle differences between preterm and term functional connectivity in both states. Lastly, we examined the relationship between connectivity, brain state and prematurity. We used simple machine learning models to examine whether brain state impacted the classification of preterm and term born infant groups. Results indicated that functional connectivity data from 2-month aged infants can classify brain state (asleep/movies) during a scan (both ex-preterm and term born infants) and can weakly predict scan age (when training on asleep state only). It is felt that methodological challenges in our analysis limited our ability to classify and predict birth gestation. Finally, study protocols are being shared with the foetal/infant/toddler neuroimaging community (FIT-NG community) which is an increasingly active and growing community of researchers. This is in the hope that both this and future work will lend towards identifying potential functional MRI biomarkers in this pre-verbal infant population. | en |
