A reliable and efficient method for cerebral ventricular volumetry in pediatric neuroimaging.

Objectives: Cerebral ventricular volume has the potential to become an important parameter in quantitative neurological diagnosis. However, no accepted methodology for routine clinical use exists to date. We sought a robust, reproducible, and fast technique to evaluate cerebral ventricular volume in young children. Methods: We describe a novel volumetric methodology to segment and visualize intracerebral fluid spaces and to quantify ventricular volumes. The method is based on broadly available T1 weighted volumetric magnetic resonance (MR) imaging, an interactive watershed transform, and a fully automated histogram analysis. We evaluated this volumetric methodology with 34 clinical volumetric MR datasets from non-sedated children (age 6-7 y) with a history of prematurity and low birth weight (= 1500 g) obtained during a prospective study. Results: The methodology, with adaptation for small ventricular size, was capable of evaluating all 34 of the pediatric datasets for cerebral ventricular volume. The method was a) robust for normal and pathological anatomy, b) reproducible, c) fast with less than five minutes for image analysis, and d) equally applicable to children and adults. Conclusions: Clinical brain ventricular volume calculations in non-sedated children can be performed using routine MR imaging besides efficient threedimensional segmentation and histogram analysis with results that are robust and reproducible.