Purpose or Case Report: Hypoxic Ischemic Encephalopathy (HIE) occurs in 1-6/1000 livebirths often leading to death or neurological disability. Neonatal brain magnetic resonance imaging (MRI) is used to assess severity of brain injury and provide prognostic information in the newborn period. However, 20-50% HIE patients do not have early abnormal MRI findings detected by qualitative neuroradiologic evaluation and yet have adverse outcomes evident by 2 years of age. Our aim was to investigate whether quantitative MRI analysis can find subtle abnormalities that help improve prognostication and reduce the incidence of falsely reassuring neonatal MRI among infants with HIE.
Methods & Materials: 28 neonates with HIE (mean gestational age=39 3/7 weeks; 11 females) with both neonatal inpatient MRIs (mean age=2.7 days) that were qualitatively interpreted as not having likely pathologic lesions and neurodevelopmental assessment data available at 2 years of age were identified by a retrospective review of health records from Boston Children’s Hospital. Patients were grouped as Normal neonatal MRI-Typical 2-year outcome (NT; n=16) or Normal neonatal MRI-Atypical 2-year outcome (NA; n=12). Atypical outcomes included developmental delay, speech delay and/or motor impairment. Apparent Diffusion Coefficient (ADC) maps for all the patients were extracted. Using our recently-developed normal neonatal ADC atlases as a reference (Ou et al., Hum Brain Mapp 2017), we automatically and quantitatively assessed any voxel-level abnormalities for individuals in the NA group. Then we tested for group-wise ADC value difference across the NT and NA group at every voxel, using the atlas as common reference, with statistical significance threshold of p<0.05 after False Positive Discovery correction.
Results: Using individual MRI analysis, atlas-based automatic lesion detection found subtle lesions in two NA individuals (17%; one with cerebral palsy and one deceased) in bilateral thalami and temporal white matter. Group analysis identified significantly lower ADC values in the NA compared to the NT group, mainly in the left thalamus (p<0.001).
Conclusions: Our preliminary results demonstrate that quantitative ADC analysis shows promise in helping to identify abnormalities in brain MRIs of HIE neonates that are difficult to characterize visually, at the individual and group level. Expanding to include analyses of structural, diffusion and tractography changes will be an important next step to optimize early prognosis by neonatal MRI among infants with HIE.