Surfacebased Atlases and Shapebased

1
Surface-based Atlases and Shape-based Analyses
of Cerebral Cortex
IPAM UCLA - July 2004
David Van Essen John Harwell Donna Hanlon James
Dickson
Washington University in St. Louis Supported by
the Human Brain Project (NIMH, NSF, NCI, NLM,
NASA) and NPACI
2
A millenium of earth cartography
1482
1630 (Hondius)
1965
2002 (Landsat Grand Canyon)
A classical atlas
3
The classical Talairach atlas (1967 1988) 3-D
coordinate system reference system/repository
4

• ATLASES of CEREBRAL CORTEX
• Central challenges
• complexity of cortical shape
• individual variability
• Issues and choices
• Visualization - surfaces and volumes
• Registration from individual to atlas
• Volume-based vs. Surface-based
• Constraints landmarks vs. continuous metric
• Coordinate systems - stereotaxic and
  surface-based

5

• OUTLINE
• Segmentation of cerebral and cerebellar cortex
• Macaque
• Human atlas
• Individual variability and its biological basis
• NEW Population-average, Landmark-
  Surface-based (PALS) atlas
• Shape-based analyses
• individual variability, hemispheric asymmetry
• Mapping fMRI population-average data to atlas
• Interspecies comparisons monkey-human

6
Surface/volume visualization Databases and web
access
Software Caret, SureFit
7
SureFit approach probabilistic maps of cortical
structure

• Outer (pial)
• boundary map
• Based on
• image intensity
• proximity to inner
• boundary

• Inner (gray-white)
• boundary map
• Based on
• image intensity
• intensity gradients

Structural MRI (input data) Noisy, low contrast
typical of conventional MRI
8
Segmented volume (Reconstruction substrate)
Segmentation on intensity image
Structural MRI (input data)
Surface reconstruction (wire-frame
tessellation) using SureFit method
9
SuMS Database http//brainmap.wustl.edu8081/sum
s/
10
(No Transcript)
11
Structural MRI
Segmentation
Surface reconstruction
12

• A mosaic of cortical areas
• Based on Function, Architecture, Connectivity,
  Topography
• Individual variability in size (3-fold for area
  V1)
• Many competing partitioning schemes! (10 mapped
  to atlas)
• Total number of cortical areas in macaque 100?

13
Lewis Van Essen (2000), Case 95D.R
14
Architectonic maps (5 hemispheres) Lewis Van
Essen (2000) partitioning scheme
80 subdivisions (areas zones) consistently
identifiable Variability in location (1-4 mm),
size, shape
15
Landmark-based registration of flat maps, using
fluid deformation algorithm of Joshi Miller
Landmarks on individual flat map (Case 95D.R,
partial hemisphere)
Landmarks on Atlas flat map
16
(No Transcript)
17
(No Transcript)
18
(No Transcript)
19
(No Transcript)
20
(No Transcript)
21
(No Transcript)
22
fMRI in monkeys Objects vs. scrambled
objects Denys et al. (J. Neurosci. 2004)
Individual (case M3)
23
Landmark-based registration of spherical maps
Individual map
Atlas map
24
Deformed fMRI (object vs. scrambled object)
Lewis-VE areas
25

• OUTLINE
• Segmentation of cerebral and cerebellar cortex
• Macaque
• Human atlas
• Individual variability and its biological basis
• NEW Population-average, Landmark-
  Surface-based (PALS) atlas
• Shape-based analyses
• individual variability, hemispheric asymmetry
• Mapping fMRI population-average data to atlas
• Interspecies comparisons monkey-human

26
(No Transcript)
27
Volumes and Surfaces after registration to
Talairach space
Buckner Case 1
Buckner Case 2
MRI data courtesy R. Buckner
28
Buckner Case 1
Buckner Case 2
Left
Left
Right
Right
29
Sulcal depth maps (Cases 1 - 6, right hemisphere)
30
(No Transcript)
31
The conundrum of cortical convolutions
Why are cortical folds so consistent in the
macaque? Why is cortical folding so complex and
variable in humans?
Complex convolutions in humans allows 10-fold
larger surface area to fit within cranium.
But what determines the pattern? Tension-based
morphogenesis?! (Van Essen - Nature, 1997)
32
(No Transcript)
33

• cortico-cortical connections
• established while convolutions form
• tension along axons causes folding?
• Van Essen - Nature (1997)

34
Tension-based Morphogenesis
35
Tension-based morphogenesis Can account for both
consistency and variability in convolutions!
Consistent folding in regions dominated by
larger areas, more robust pathways
Variability in 'balkanized' regions (many small
areas) outcomes influence by genetic,
developmental factors.
36
Implications for registering individuals to
atlas Macaque - all major sulci are reliable 17
landmark contours Human - only a few sulci are
reliable. Corresponding sulcal names does not
imply corresponding functional
organization. Less is more! 6 'core' landmark
contours
37
(No Transcript)
38

• OUTLINE
• Segmentation of cerebral and cerebellar cortex
• Macaque
• Human atlas
• Individual variability and its biological basis
• NEW Population-average, Landmark-
  Surface-based (PALS) atlas
• Shape-based analyses
• individual variability, hemispheric asymmetry
• Mapping fMRI population-average data to atlas
• Interspecies comparisons monkey-human

39
(No Transcript)
40
(No Transcript)
41

• OUTLINE
• Segmentation of cerebral and cerebellar cortex
• Macaque
• Human atlas
• Individual variability and its biological basis
• NEW Population-average, Landmark-
  Surface-based (PALS) atlas
• Shape-based analyses
• individual variability, hemispheric asymmetry
• Mapping fMRI population-average data to atlas
• Interspecies comparisons monkey-human

42
(No Transcript)
43
Human Visuotopic (fMRI) and orbitofrontal
(architectonic) Mapped via surface-based
registration to atlas
44
Mapping population-average fMRI volume data to
atlas surfaces Its easy, but subject to
biases of target surface
45

• OUTLINE
• Segmentation of cerebral and cerebellar cortex
• Macaque
• Human atlas
• Individual variability and its biological basis
• NEW Population-average, Landmark-
  Surface-based (PALS) atlas
• Shape-based analyses
• individual variability, hemispheric asymmetry
• Mapping fMRI population-average data to atlas
• Interspecies comparisons monkey-human

46
Human fMRI - Objects vs. scrambled
objects (population average, volume-registration
same stimuli/paradigm as in macaque)
K. Denys, G. Orban, W. Vanduffel, D. Van Essen
(unpublished)
47
Monkey vs. Human - comparisons on same task
48
Comparisons across species

• What areas/regions are homologous?
• Evolutionary divergences - where and how?
• Test hypotheses using surface-based registration

49
(No Transcript)
50
(No Transcript)
51
(No Transcript)
52

• OUTLINE
• Segmentation of cerebral and cerebellar cortex
• Macaque
• Human atlas
• Individual variability and its biological basis
• NEW Population-average, Landmark-
  Surface-based (PALS) atlas
• Shape-based analyses
• individual variability, hemispheric asymmetry
• Mapping fMRI population-average data to atlas
• Interspecies comparisons monkey-human
• Databases - coping with the data flood!

53
(No Transcript)
54
Atlases (human, macaque, mouse), Caret
visualization software freely available
(http//brainmap.wustl.edu)