analysis using surface-based methods
Traditionally, in the analyses of human brain structure, region
of interest (ROI) methods have been used to suggest volumetric
changes and differences within/between groups. Manually segmented
ROI-based analyses are intensive labor and suffer from the issues
of intra- and interrater reliability issues and different ROI
definitions. Voxel-based morphometry (VBM) is an observer-independent
automatic methodology that is implemented across the cortex. However,
this method also exhibits limited accuracy in the measurement
of cortical morphology, especially in brain regions where fine
anatomic details are often obscured by a partial volume effect.
The concepts of gray matter density in VBM are central to the
interpretation of the results of group data. Recently, to reflect
more accurate cortical geometry and overcome partial volume effect,
the analyses using cortical surface model have been performed.
Cortical surface-based analysis can make it possible to measure
cortical thickness in millimeters (more precise measurement in
deep sulci as a cortical sheet). In addition, cortical shape and
folding pattern can be analyzed, which is not possible in volumetric
The native MR images are processed through the anatomical pipeline.
The surfaces of the inner and outer cortex were automatically
fitted using the Constrained Laplacian-Based Automated Segmentation
with Proximities algorithm (Kim et al., 2005, NeuroImage) that
is sort of deformable model algorithm. The reconstructed hemispheric
cortical surfaces consisted of 81,920 high-resolution meshes
of discrete triangular elements.
measurements for cortical structure analysis
The inner and outer surfaces had the same number of vertices,
and there was a close correspondence between the counterpart
vertices of the inner and outer cortical surfaces. The cortical
thickness was defined as the Euclidean distance between these
Surface-based diffusion smoothing and 2-D spherical surface
registration are performed to vertex-by-vertex analysis.
Our application study
- Gender difference in cortical thickness (Im et al., 2006,
- Cortical thickness in MCI (Seo et al., 2007, NeuroImage)
Cortical surface area is calculated, which is the straightforward
sum of the areas of the triangles making up the surface model.
Cortical surface area can be used to suggest the overall degree
Previous studies have reported that the change of sulcal depth
is associated with brain disease. Sulcal depth maps are generated
by measuring the 3-D Euclidean distance from each vertex in
the cortical surface to the nearest voxel on a cerebral hull
volume. The Euclidean distance as the measure of sulcal depth
was used in previous studies and one of them showed cortical
folding abnormalities in Williams Syndrome.
convolution (mean curvature)
In order to quantify the degree of sharpness and convolution
of cortical folding regardless of cortical depth, mean curvature
was computed at the vertices that lie within the cortical surface.
Mean curvature on sulcal walls is measured to estimates the
degree of sulcal convolution regardless of the effect of sulcal
fundic sharpness in order to differentiate between these scenarios
for more definite interpretation. Areas of sulcal wall are defined
using the sulcal depth map as lying between specific depths.
Fractal dimension (FD) provides a quantitative description of
the structural complexity in the cerebral cortex. After extracting
3D vertex information of cortical surface, FD is measured using
Our previous study (Im et al, 2006, Hum. Brain Mapp.)
FD is an extremely compact measure of shape complexity, condensing
all details into a single numeric value. We interpreted the
variation of the FD in the cortical surface of normal controls
through multiple regression analysis with cortical thickness,
sulcal depth and folding area, related to cortical complexity.
Human cortex develops a complex structure through the thinning
of cortical thickness and by increasing the frequency of folds
and the convolution of gyral shape rather than by deepening
Folding area calculates the area of the folded regions normalized
by whole surface area in the outer smoothed cortical surface
model. The purpose is to detect how large area the folding regions
occupy in entire surface area. The folding regions are defined
using mean curvature in the 3-D triangular surface model.
Our Application study using folding area and mean curvature
(Im et al., 2007 in OHBM conference)
The purpose is to analyze the pattern of cortical folding in
MCI and AD. Using the combination of folding area and mean curvature,
the results suggest that sulcal widening becomes more severe
across the entire cortex in AD than in old normal and MCI groups.