Functional MRI of the developing neonatal brain: potential and challenges for the future

Abstract

This commentary is on the original article by Lee et al. on pages 724-729 of this issue In the 20 years following its description by Ogawa et al., blood oxygen level-dependent (BOLD) contrast functional magnetic resonance imaging (fMRI) has become established as the neuroimaging tool of choice for neuroscience and psychology studies of the brain.1 The technique offers both relatively good spatial and temporal sensitivity, with whole brain visualization a major advantage, in contrast to other tools such as electrophysiology and near infrared. Moreover, the quantitative nature of the signal, relative experimental flexibility, and reproducible dynamic signal properties across different infants have made it ideal to answer detailed neuroscientific hypotheses with the application of higher level statistical testing and mathematical modelling. This has allowed investigators to map and characterize the functional organization of the brain accurately, and begin to probe more complex issues such as the underlying patterns of connectivity between different regions during particular tasks and at rest.2 These attributes, coupled with its non-invasive nature, would seemingly make fMRI extremely suitable for studying the activity of the developing human brain. Despite this apparent potential, there have only been a small number of fMRI studies in neonatal infants which have reported reliable functional responses to simple visual, auditory, and tactile stimuli.3 However, within these studies there has been prominent inconsistency as to the amplitude of the functional responses, with early studies reporting (in contrast to the typical adult response) a localized decrease in BOLD signal (termed ‘negative’ BOLD). This variation is likely to be due to the considerable challenges inherent in designing and performing an fMRI study for neonatal infants, as it is generally inappropriate to directly apply existing techniques developed for adult studies. These challenges include making necessary adaptations to the MRI acquisition parameters to account for the different magnetic properties of the developing brain; the design of appropriate and safe stimulation paradigms; understanding the effects of physiological changes on the underlying biophysics of the BOLD contrast response; and making adjustments to the analysis techniques.4 Lee et al. have performed a feasibility study of fMRI using a visual stimulus in a relatively large cohort of preterm infants.5 They have attempted to address some of the possible sources of irregularity described in previous studies, in particular the considerably longer T2* values seen in neonatal infants predominately due to the reduced synaptic density and increased water content of the immature brain.4 Despite this, functional responses were seen in just 5% of their preterm infants, suggesting that fMRI is probably not an appropriate modality for studying visual function in this population. Of note, a visual stimulation paradigm was used in the majority of previously reported neonatal fMRI studies, despite the greatest inconsistency in results occurring with this stimulus type.3 Although newborn infants behaviourally are visually attentive (e.g. showing a preference for looking at images of faces) and immature visual evoked potentials can be demonstrated in some infants as young as 24 post-menstrual weeks, the study’s findings may be in keeping with the theory that at least some early visual activity occurs via the subcortical extra-geniculo-calcarine system; with functional activity in the primary visual areas not fully established until 2 to 3 months of age.6 Although recent neonatal fMRI studies have begun to report more consistent results (and in particular positive BOLD signal responses) with auditory and sensorimotor stimuli,7,8 there remains a number of key unresolved questions; such as the effect of developmental changes on the local neurovascular coupling which underlies the BOLD response and has been shown to differ significantly in the immature rat brain.4,9 If some of these issues can be definitively resolved, then one could imagine fMRI providing a wealth of new information about the development, organization, and integration of functional activity in the newborn brain, and also important insights into the pathophysiology of central nervous system developmental abnormalities.