RATIONALE AND OBJECTIVES: The aim of this study was to assess the ability of a semiautomated process to produce three-dimensional reconstructions of the ventricles and calculate ventricular volumes from magnetic
resonance (MR) imaging data in children with structural
brain abnormalities. MATERIALS AND METHODS: Fourteen children referred for MR imaging of
the brain for
neurologic symptoms were
selected. Seven participants had structural
brain abnormalities on MR imaging; seven further participants were age-matched controls with normal
brain morphology. MR imaging included T1-weighted volumetric images in all cases. Semiautomated postprocessing techniques were performed on the MR imaging data to generate three-dimensional reconstructions of the ventricles. These were analyzed for morphologic changes, and volumes were calculated. Inter- and intrarater agreement of ventricular volumes were calculated. RESULTS: This technique produced detailed three-dimensional reconstructions of the ventricles, even in children with grossly abnormal ventricular morphology. All MR imaging data were successfully postprocessed in <5 minutes. Inter- and intrarater reliability was excellent, with
correlation coefficients of 0.99 and 0.92, respectively. CONCLUSION: This methodology can create detailed three-dimensional visualizations and volumetric measurements of morphologically abnormal ventricles. This technique could help physicians and parents comprehend abnormal ventricular
anatomy better and may have future clinical uses in monitoring disease progression or
neurosurgical planning.
