Monday 26 Mar 2018: Dynamics Seminar: Connectomics and beyond - Markers of cognitive training and disease
Markus Schirmer - Harvard
LSI Seminar Room A 14:30-15:30
One of the most interesting areas of investigation using neuroimage analysis is finding markers that reflect changes in the brain. These changes may be due to cognitive training or the result of an underlying disease. In recent years, network analysis in the brain (connectomics) has gained much interest, as it helps us describe the interaction of brain regions with another in a principled manner. Recently, heat- kernels have been shown to be a sensitive measure of global information transport, allowing us to differentiate patient groups from healthy controls. In the first part of this talk I present our recent work on fluid intelligence training, as well as autism, and demonstrate the efficacy of heat-kernel signatures to describe changes in the brain in these cohorts. To go beyond connectomics, we have to consider the structural elements that underly brain connectivity. In case of structural connectivity, it is the underlying health of the white matter that is crucial. However, in many diseases, we can observe so called white matter hyperintensities, whose manual characterization is labor intensive and poses a challenge in big data analyses. The second part of my talk therefore focuses on the automated differentiation of healthy and diseased white matter in stroke patients, which will help pave the way for more accurate disease descriptors that can combine general neuroimage markers and connectomics.