Valdis Gudmundsdottir PhD Student is currently employed as a Marie Curie doctoral fellow (EVAN Training Network). She is a PhD student working on the applications of paleoanthropological methods for the study of the human cortex under the scientific direction of Jean-Francois Mangin.
The objective of this thesis is the transfer of morphometric methods from paleo-anthropology towards neuroimaging.
One of the key differeces between the two fields of study is the
number of specimens one works with. A paleo-anthropologist will
typically work with relatively few specimens while in neuroscience a
database can easily comprise 100 brains. The paleo-anthropologist can
place the landmarks manually on the surface of their specimen while in
neuroimaging a method must be devised to place landmarks automatically
on the cortex.
The host institution, Neurospin, has developped a method to
automatically recognise the cortical folds. See figure 1 (a-b). Their
extremities and centres of gravity are then calculated and used as
landmarks for a geometric morphometric analysis. See figure 1 (c-d).
The ensemble of landmarks for 62 subjects that are submitted to the shape analysis can be seen in the figure 1 (e-f).
Using these landmarks, we have compared the shape of the right
and left hemispheres and have been able to identify the large mode of
asymmetry in the human cortex where due to language being commonly
lateralised to the left side of the brain, the posterior end of the
sylvian fissure is shifted backwards in the left hemisphere. We
validated the method on a manually labelled dataset seen on the left in
figure 2 and then applied the method to a completely automatically
labelled dataset, seen on the right hand side of figure 2. The arrows
show well the global asymmetry effect at the posterior end of the
Having validated the method on a known shape effect in the brain we
will now apply the method to look for early signs of psychiatric