Although it has been shown that MRI provides more comprehensive information of the neonatal brain structure, including more subtle abnormalities unable to be detected on cUS, the clinical implications of these findings require standardization, validation and long term follow-up. There have been numerous studies that have evaluated the ability of MRI at term equivalent age to predict neurodevelopmental outcomes at 12 months to 9 years. The studies with more recent cohorts, large sample sizes and quantitative predictive data are summarized in Table 2.

White Matter Abnormalities

The most extensively studied MRI abnormality shown to predict neurodevelopmental outcomes is white matter lesions. Types of white matter abnormalities have included periventricular leukomalacia, punctate white matter lesions (PWML), and Diffuse Excessive High Signal Intensities (DEHSI). The etiology of these entities has yet to be completely understood. Specifically in regards to PVL and PWML, it is unclear if they represent a spectrum of one disorder or separate entities with different etiologies, but they both correlate with long-term neurodevelopmental outcomes. In contrast, DEHSI has been shown in numerous studies to not correlate with long-term outcomes.2 Periventricular leukomalacia is a form of white matter injury, which is focal and cystic. It is histologically characterized by coagulative necrosis within the subventricular lamina of the periventricular white matter, leading to the destruction of cellular elements and the formation of cysts, and possibly secondary ventriculomegaly Punctate white matter lesions are non-specific focal lesions in the white matter on MRI that are also commonly found in preterm infants. They are represented as increased signal on T1-weighted imaging and decreased signal on T2-weighted imaging and may histologically reflect petechial hemorrhages, gliotic scarring, or mineralization. Diffuse excessive high signal intensities (DEHSI) are a frequent finding on T2-weighted MR images near-term in children born prematurely3 – seen in up to 80% of preterm infants at term equivalent age.4,5 They are hyperintense white matter signal abnormalities found in the periventricular and subcortical white matter regions when compared to the signal intensity of normal unmyelinated white matter. The most common outcome measure in studies that look to correlate MRI findings and neurodevelopment outcomes is motor dysfunction, ranging from mild motor delay to severe cerebral palsy. Although the ability of MRI to predict cerebral palsy can be variable, one of the most important observations from past studies is that preterm children without white matter abnormalities on term MRI did not show significant neurodevelopmental deficits compared to full term controls – specifically in regards to motor 6 cognitif and executive functioning 7,8 demonstrating that MRI has a high negative predictive value. Compared to infants born at term, preterm infants though have been shown to have cognitive impairments, requiring special assistance at school. White matter abnormalities have been shown in studies to be associated with lower cognition, language abilities, and executive functioning.

Gray Matter Abnormalities

Gray matter can be assessed for both morphology, including cortical folding and surface area, as well as signal abnormalities. Gray matter abnormalities (assessed by size of the subarachnoid space, quality of gyral maturation and presence of gray matter cortical signal abnormality) are also associated with poorer neurodevelopment outcome.

Cerebellar Abnormalities

Cerebellar abnormalities have been described as destructive lesions vs. those representing impaired development. Destructive lesions include cerebellar hemorrhage and infarction. Cerebellar hemorrhage is usually unilateral and associated with supratentorial lesions, leading to possible cerebellar atrophy with time 9 which is related to neurodevelopmental sequelae.33

There have been different classification systems proposed and used for the interpretation of MRI data defining MR abnormalities and relate them to neurodevelopmental outcome (refer to Table 3) Woodward et al described and used a comprehensive scoring system to define brain abnormalities in preterm infants near term age ..11,12 A separate scoring system was used to define the extent of gray matter (GM) injuries. Moderate-to-severe cerebral white-matter abnormalities were predictive of cognitive delay with odds ratio (OR) of 3.6, motor delay with an OR of 10.3, and CP with an OR of 9.6. The OR of gray matter abnormalities in association with severe cognitive delay or CP ranged from 2-3.11 This scoring system has been further refined by Kidikoro et al.13{These scoring systems offer a more objective and comprehensive method to classify magnitude of brain injury in the preterm population given the wide range of complex MRI abnormalities associated with prematurity, as well as predict neurodevelopmental outcomes. In recent neuroprotective drug trials these methods have helped to monitor drug intervention effects 14

Timing of MRI

Both qualitative and quantitative MR techniques have been used to assess the preterm brain at different stages in development, including in the perinatal period, at term equivalent age, and through childhood into adolescence. Compared to the MRI at time of birth, the MRI at term equivalent age still remains the optimal scanning time for children born prematurely in the prediction of neurodevelopmental outcomes, given the greater clinical stability of the population at term as well as the success of feed and wrap protocols in obviating the need for sedation. In addition, the most commonly studied timing of MRI scans in the preterm population to predict neurodevelopmental outcomes is term equivalent age. However, there have been some studies investigating the correlation between serial MR imaging and neurodevelopmental outcomes, looking at changes in findings over time.

The development of sophisticated new MRI strategies has permitted a better understanding of corticogenesis in the prematurely born. Furthermore, data from many of these strategies, including volumetric imaging, diffusion tensor imaging, MRS and functional connectivity, has been shown to correlate with cognitive measures in preterm subjects at school age, adolescence and young adulthood. Below we describe these strategies, and the published neonatal outcome data available. Although these studies do not represent the predictive utility of volumetric MRI in terms of PPV/NPV or sensitivity/specificity, the outcome data generated from each is presented in Table 4.

Volumetric Imaging

Volumetric analysis uses two or three-dimensional MR images to quantitatively characterize alterations of brain development associated with prematurity and brain injury. It allows for the measurement of volume of specific brain structures including the cortical and subcortical regions, cerebellum and hippocampus. Automated segmentation techniques can be used to differentiate grey matter and white matter as well as unmyelinated and myelinated tissue.15 Volume reduction in the whole brain, anatomic structures in the brain, and regionally within white matter and cortical grey matter have been reported in preterm infantsIn addition, simple brain metric measurements using MRI as a method to assess brain growth and atrophy in the neonates has been shown to correlate with outcome Smaller brain volume at term equivalent age, when correcting for effects of white matter injury, is correlated with subsequent performance on object working memory tasks in infancy.16 Subsequent working memory performance in infancy was also shown to be associated with reductions in volumes of dorsolateral prefrontal, sensory, motor, parieto-occipital and premotor cortex.16 Total white matter volumes in sensory, motor and mid-temporal regions at term equivalent age are strong predictors of neurodevelopmental outcome in the first year of life.17 Inder et al18 showed that the overall gray matter volume at term is predictive of neurodevelopmental outcome, while Boardman showed association between decreased volumes of deep gray matter and neurocognitive outcome.19

Diffusion Tensor Imaging

Assessing the microstructural development of the white and gray matter in prematurely born infants has become feasible with the recent advantages in diffusion imaging. Inferring information about the underlying tissue microstructure from dMRI is an inverse problem, were models of different complexity are fitted to the acquired signal . The most popular model to describe changes in the dMRI signal intensities is the diffusion tensor model (DTI). In this model, the directional motion of water behavior is characterized by a 3 dimensional tensor. Diffusion tensor imaging measures relies on the water motion providing the useful parameters for describing the underlying microstructure. Measure of fractional anisotropy (FA) describes the degree to which water diffusion is restricted in one direction relative to all others. FA values of major white matter tracts were correlated with gross and fine motor performance. Lower FA in regions of corpus callosum, at term equivalent age, was shown to correlate with the psychomotor developmental index Furthermore, the lower values of FA at term in posterior limb of internal capsule were associated with neurodevelopmental outcomes in first year of life. Follow-up of children with abnormal neurological exams in the first year of life showed decreased FA values in posterior limb of the internal capsule at term age when measured retrospectively. In recent years DTI has been extended to use multiple gradient directions to assess better the directionality of water diffusion in crossing fibers and multiple b-values to assess diffusion characteristics in tissue compartments defined by biophysical models such as NODDI