Biomedical Engineering Reference
In-Depth Information
FIGURE 14.2
Flow diagram of the ICBM
1
pipeline. All MRI data is processed through the pipeline shown
above. After preprocessing and stereotaxic registration, the cortical surface is extracted and
the MRI data is classified into gray matter, white matter, and cerebrospinal (CSF) components.
FIGURE 14.3
Average T1-weighted MRIs of, from left to right, a pediatric population (
n
130), healthy
adults (
n
151), and MS patients (
n
460).
FIGURE 14.4
Flow diagram of the INSECT
73
pipeline.
After the data have been registered and resampled into stereotaxic space,
they are available for processing. As a first step, voxel-by-voxel intensity aver-
ages can be computed for each group studied. Figure 14.3 shows a trans-
verse slice through three different group averages. While useful as a
qualitative indicator of local anatomical variability for the different groups,
the composite MRI intensity averages are insufficient as a quantitative tool.
For this purpose, the MRI intensity for each voxel must be identified with a
tissue label or structure label. The former requires automated classification,
while the latter requires segmentation. Figure 14.4 shows an incarnation of
INSECT (intensity normalized stereotaxic environment for classification of
tissues
73
), a pipeline designed for the classification of tissues in MRI. This
pipeline was applied to the 152 brain volumes in the ICBM data base, clas-
sifying each voxel into gray matter, white matter, and cerebrospinal fluid.
Figure 14.5 shows statistical probability anatomy maps (SPAMs) that were
