Nancy K. Rollins1,2, Youngseob Seo1,2, Lina Chalak1,2, Jonathan M. Chia3, Gareth Ball4, Zhiyue J. Wang1,2
1University of Texas Southwestern Medical Center, Dallas, TX, United States; 2Children's Medical Center, Dallas, TX, United States; 3Philips Healthcare, Cleveland, OH, United States; 4Imperial College and MRC Clinical Science Center, London, United Kingdom
Diffusion tensor imaging (DTI) is being used to study brain maturation and disease effects on white matter. Manual region-of-interest (ROI) analysis which is widely used is operator-dependant, tedious, hypothesis driven, and impractical for comparison of groups or large numbers of subjects but does not require potentially confounding spatial normalization. Tract Based Spatial Statistics (TBSS) is an open source operator-independent platform which spatially normalizes tensor data; potentially problematic in neonates as the normalization templates used in adults have not been proven reliable to infants. We performed a direct comparison between manual ROI analysis and TBSS in infants using the modifications proposed by Ball to improve the reliability of TBSS in neonates and then report region-specific tensor metrics in infants 32 -49 weeks post conception age at the time of imaging.