Biomedical Engineering Reference
In-Depth Information
FIGURE 14.1
Example of a single-subject pipeline verification image, obtained from analysis of the brain
of a healthy child. Each column shows a different slice through stereotaxic (Talairach) space,
while each row shows different intermediate and final products of the automatic analysis.
Top row: T1-weighted volume with several contours, showing the axes of Talairach space,
the outline of the MNI 305 average brain, and a low- and high-resolution cerebral surface.
78
Second row: 3-class INSECT classification.
73
Third row: smoothed (FWHM
10) gray matter
classification.
81
Fourth row: T1-weighted image, nonlinearly deformed at low resolution to
match a model brain, shown with its cerebral and ventricular contours.
After running through this basic pipeline, a subject's MRI volume can be
visualized in stereotaxic space with its corresponding tissue labels, anatomi-
cal structure labels, cortical surface, and sulcal ribbons—all in 3D. As a stan-
dard procedure, a number of composite verification images are produced
during processing to allow the rapid visual inspection of the results for a
large number of data sets (see Figure 14.1).
The following sections describe the application of pipeline processing in
cohort studies for the analysis of brain data from (1) a collection of normal
young adults within the ICBM project, (2) an ensemble of children five to eigh-
teen years of age, and (3) a group of patients suffering from multiple sclerosis.
14.3.1
Registration-Based Analysis of Normal Brain Anatomy
Figure 14.2 shows the pipeline designed for the automatic analysis of normal
human brain, which has so far been applied to brain MRIs (T1-weighted 3D
spoiled gradient echo acquisition with sagittal volume excitation, TR
18,
TE
30°, 140-180 sagittal slices) obtained from 152 healthy adults
(86 male, 66 female, age 24.6
10, flip angle
4.8)
1,82
and 111 healthy children and adolescents.
83
