[BIC-announce] Measuring Cortical Thickness from MRI - (Pre-defence PhD Thesis Seminar by Jason LERCH)

[Reza ADALAT]

Date & Time: Friday, April 8th. at 1:00 PM.
Place: de Grandpre Communication Center.

Everyone is welcome.


Abstract:

The thickness of the human cerebral cortex, defined as the geometric
distance between the white matter to grey matter boundary and the pial
surface, is a fundamental property of the cortex and amenable to
investigation from in-vivo Magnetic Resonance Imaging (MRI). The aim of
my PhD was to evaluate different cortical thickness metrics, to assess
the contributions cortical thickness measurements can make to the study
of Alzheimer^?s Disease (AD), and to explore statistical techniques for
analyzing cortical thickness.
Evaluation of different thickness metrics was carried out within the
context of a population simulation. 50 young normal subjects were
chosen, and the cortex artificially thinned in an isolated cortical
patch in half the population. The cortices were then extracted on all
50 subjects, and thickness measured using six different metrics. One
method, t-link, had the highest precision with a sensitivity of 93%.
Moreover, the use of surface-based blurring was assessed along with
optimal techniques for thresholding and correcting the results for
multiple comparisons.
The usefulness of cortical thickness to real data was assessed in the
context of AD. 19 patients and 17 controls acquired in Munich, Germany,
had their MRI scans processed to obtain cortical thickness. Striking
and highly significant differences were found between the two groups.
The greatest effect occurred in the medial temporal lobes, the putative
origin of AD pathology. These changes were further related to disease
progression.
The final part of this PhD consisted of testing alternate statistical
methods for analyzing thickness data. The first such application was to
test different discriminant techniques to assess whether cortical
thickness could prove an aid in the diagnosis of AD. Using additive
models containing the mean thickness of cortical structures resulted in
an accuracy of 95%, and showed quadratic discriminants to be the most
effective.
In a similar vein, a large normal childhood development cohort was used
to test how the thickness of different areas of the cortex relate to
each other, and whether these relationships provide additional insight
into, for example, normal development or the relationship between
cortical thickness and IQ. The results showed that the association
cortices are most correlated with the rest of the cortex, that
correlations Brodmann Area 44 and the superior temporal gyrus tighten
with age, and that subjects with higher IQ showed increased correlation
between different frontal and parietal regions than low IQ subjects.
Cross-cortical correlations should thus be considered an additional
method of analyzing population data using cortical thickness.
The four studies included in this PhD have further entrenched cortical
thickness measurements from MRI as a valuable technique for studying
the anatomy of the cortex. Collaborations with researchers both inside
and outside of McGill have applied the methods developed herein to
congenital amusia, Schizophrenia, multiple sclerosis, epilepsy, normal
ageing, ADHD, and Kleinfelter^?s Disease. The use of cortical thickness
is thus going to be of increasing importance in the neurosciences over
the coming years.