Diagnosing pediatric encephalitis is challenging because of varied clinical presentation, nonspecific neuroimaging features, and rare confirmation of causality. We reviewed acute neuroimaging of children with clinically suspected encephalitis to identify findings that may correlate with etiology and length of stay.

Imaging of 141 children with clinically suspected encephalitis as part of The California Encephalitis Project from 2005 to 2012 at a single institution was reviewed to compare the extent of neuroimaging abnormalities to patient age, gender, length of stay, and unknown, possible, or confirmed pathogen. Scan review was blinded and categorized by extent and distribution of abnormal findings.

Abnormal findings were evident on 23% (22/94) of computed tomography and 50% (67/134) of magnetic resonance imaging studies in the acute setting. Twenty children with normal admission computed tomography had abnormal findings on magnetic resonance imaging performed within 2 days. Length of stay was significantly longer among children with abnormal acute magnetic resonance imaging (P < 0.001) and correlated with increased complexity (Spearman rho = 0.4, P < 0.001) categorized as: no imaging abnormality, meningeal enhancement and/or focal nonenhancing lesion, multifocal lesions, confluent lesions, and lesions plus diffusion restriction, hemorrhage, or hydrocephalus. There was no correlation between neuroimaging findings and an identifiable pathogen (P = 0.8).

Abnormal magnetic resonance imaging findings are more common than abnormal computed tomography findings in pediatric encephalitis. Increasing complexity of magnetic resonance imaging findings correlated with disease severity as evidenced by longer length of stay, but were not specific for an identifiable pathogen using a standardized diagnostic encephalitis panel.