|
Guido Gerig, PhD Imaging studies of early brain development get increasing attention as improved modeling of the pattern of normal development might lead to a better understanding of origin, timing and nature of morphologic differences in neurodevelopmental disorders and brain disease. UNC has collected one of the largest samples of longitudinal MRI/DTI of neonates and infants up to age 5 years (N>300) to model the trajectory of early brain development using structural MRI (sMRI) and diffusion tensor imaging (DTI). Studying MRI of this age group involves two major challenges, successful high quality MRI scanning of non-sedated infants and image analysis methods that can cope with very low contrast-to-noise ratio, small brain size, variability of brain shape and size, and locally varying contrast changes due to a continuous change of early myelination. We will discuss the challenge of scanning children in this age group but then focus on image analysis aspects, since standard methodology developed for adult MRI is inadequate for images of this age group This will include new tissue segmentation for partially myelinated white matter, parcellation of full brain into major lobar and subcortical structures, and computational anatomy tools for building age specific atlases and for modeling local growth throughout the whole volume. Shape analysis of multi-object complexes, here the set of subcortical structures, is demonstrated to describe major changes due to longitudinal growth and cross-sectional group differences. White matter development is assessed by novel region-based statistical analysis of diffusion tensor imaging (DTI) and tools for quantitative analysis of major fiber tracts. We will present latest results on lateralization, gender differences, very early trajectory of cortical gray matter growth, and the development of white matter as seen most rapidly in the first year of development but still very significant between 2 and 4 years. Measuring the early trajectory of growth via structural MRI and DTI will likely provide a vastly improved understanding of early brain development, of changes due to delayed development or pathology, and of its relationship to neuropsychiatric disorders. Biosketch Guido Gerig is a Taylor Grandy Professor of Computer Science and Psychiatry at the University of North Carolina at Chapel Hill, where he studies methods for segmentation and visualization of multidimensional medical data as well as the statistical analysis of 3D shape. He holds a MSc. in Natural Sciences, and both a Ph.D. and a PD in Electrical Engineering and Computer Science all from ETH Zurich . Guido serves on several program committees and editorial boards for medical image analysis journals and conferences, and he was awarded the UNC Computer Science Students Association Teaching Award in 1999. |