Sponsor: Prof. Sidney Spector

1
Sponsor Prof. Sidney Spector Computational
anatomy to assess growth pattern of early brain
development in healthy and disease populations
Guido Gerig1, 2 John H. Gilmore1 Matthieu
Jomier1 Sarang Joshi3, 2 Joseph Piven1
Departments of 1Psychiatry, 2Computer Science,
3Radiation Oncology University of North Carolina,
Chapel Hill,NC 27614, USA, gerig_at_cs.unc.edu,
http//www.cs.unc.edu/gerig
Background
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. Quantitative MR imaging studies face
the challenge that cross-sectional
inter-individual variability is very large in
relation to longitudinal change, which
underscores the critical importance of a
longitudinal design of such studies. It is our
goal to model the trajectory of early brain
development, primarily focusing on the most
challenging group of very young children in the
age range from birth to 6 years, as a
4-dimensional atlas that is represented by a time
series of 3-D images and quantitative description
of local growth. In addition, the same technique
is applied to generate representative atlases for
various groups, e.g. group-specific atlases for
female/male populations and for healthy controls
and patients.
Adult
Trajectory of Brain Growth
Brain growth in longitudinal study (Aut/DD/Typ)
of children at 2yrs and 4yrs. Notice the large
cross-sectional variability.

• Pediatric neuroimaging studies at UNC
  contributing to this research
• Longitudinal study of neonatal brain development
  in high risk children and controls (Ntot134),
  follow-up at 1yr (PI John H. Gilmore).
• Autism study with baseline scans at 2yrs and
  follow-up at 4yrs (51 AUT, 25 TYP/DD)

• Open Issues
• Modeling growth trajectory of early brain growth,
  normative data of average brain and variability
• Quantitative description of local brain changes
  throughout the whole volume
• Assessment of group differences and hypothesis
  testing directly on volumetric image data

Methods

• Computational Anatomy Tool
• Unbiased atlas building by simultaneous
  diffeomorphic high-dimensional deformation of a
  population of images
• Group difference and longitudinal change analysis
  by describing volumetric deformation between
  atlases
• Differentiation of deformation field (Jacobian)
  describes local volume changes
• New tool is not limited to the cortex but extends
  to a fully volumetric description of
  changes/differences, including white matter,
  cortical gray, subcortical structures and csf.

ATLAS Neonates
Sarang Joshi, Brad Davis, Matthieu Jomier, Guido
Gerig, Unbiased Diffeomorphic Atlas Construction
for Computational Anatomy, vol. 23, NeuroImage
2004