Download ACNP05_2_Gerig_CompAnat

Preliminary Results Longitudinal Change
Average 2yrs
Atlas
Average 4yrs
Atlas
Yr 2
Yr 4
Log|Jac|
Brain surface illustration of local brain growth between
2yrs and 4yrs. Growth is most apparent in right frontal and
left occipital cortex and in general in cortical gray.
growth
Preliminary Results Group Tests
The most striking result of the
longitudinal growth analysis between 2
and 4 years is the apparent cerebral
asymmetry and brain torque. There is a
consistent right frontal > left frontal and
a left posterior parietal/occipital > right
posterior parietal/occipital pattern,
commonly called torque or brain
torsion. The temporal lobes show a
similar pattern as the frontal lobes, with
right temporal > left temporal growth.
Local growth is mostly evident in
cortical gray, which seems to account
for the major brain growth. Lateral
ventricles are stable, but the third
ventricle illustrates a significant width
reduction, along with a closing of the
ascending ramus of the Sylvian fissure.
shrinkage
Cross-sectional group difference analysis between Typ and Aut
Yr 2
Autism
2yrs
Typical
2yrs
Yr 4
Autism
4yrs
Typical
4yrs
Design of cross-sectional and longitudinal
group tests. Atlases are created for the
autism and control groups at age 2 and 4yrs.
We used 5 images in each group for this
preliminary feasibility study.
Cerebellum differences
Typ > Aut
Typ < Aut
Group tests between TYP and AUT subjects at age 2 (left) and
age 4 (right) reveal a strong group difference of the cerebellum,
which is much more pronounced at age 2 and lessens towards
age 4.
Conclusions
Our preliminary findings indicate that the new methodology shows excellent potential to explore longitudinal
change, difference between groups, and differences between growth trajectories between groups. The
simultaneous analysis of the whole volumetric brain is a major strength, as it will reveal morphometric changes of
structures with embedding context, e.g. studying cortical growth in relationship to adjacent white matter, and
examining groups of subcortical structures and even whole circuits. We are currently confirming these exploratory
findings with larger samples. Gender differences could not yet be explored due to the small sample size.
Literature
• Sarang Joshi, Brad Davis, Matthieu Jomier, and Guido Gerig, "Unbiased Diffeomorphic Atlas Construction for Computational Anatomy,"
NeuroImage; Supplement issue on Mathematics in Brain Imaging, (PM Thompson, MI Miller, T Ratnanather, RA Poldrack, and TE Nichols, eds.),
vol. 23, no. Supplement1, pp. S151-S160, Elsevier, Inc, 2004.
• Peter Lorenzen, Marcel Prastawa, Brad Davis, Guido Gerig, Elizabeth Bullitt, and Sarang Joshi, Multi-Modal Image Set Registration and Atlas
Formation, Medical Image Analysis MEDIA, Elsevier, in press
• Peter Lorenzen, Brad Davis, Guido Gerig, Elizabeth Bullitt and Sarang Joshi, "Multi-class Posterior Atlas Formation via Unbiased Kullback-Leibler
Template Estimation", Lecture Notes in Computer Science LNCS 3216, Springer Verlag, pp. 95-102, 2004.
• B Davis, P Lorenzen, and S Joshi, "Large Deformation Minimum Mean Squared Error Template Estimation for Computational Anatomy," in
International Symposium on Biomedical Imaging (ISBI) Proceedings, (Piscataway, NJ), pp. 173-176, Apr. 2004.
Acknowledgements
Supported by the NIH Conte Center MH064065, the Neurodevelopmental Disorders Research Center HD 03110,
the NIH RO1 MH61696, and NIBIB grant P01 EB002779