Science

The goal of this project is the development of local models for fibre orientation distribution, accounting for crossing and fanning structures. Furthermore, we work on effective deterministic and probabilistic fibre tracking algorithms. In order to improve the applicability of the techniques, we realize speed effective implementation using parallelization. Systematic comparisons are performed for a broad selection of current state-of-the-art fibre tracking algorithms. We apply these techniques to to various cognitive systems. [more]

In this project we develop techniques for the parcellation of the cortical sheet on the basis of probabilistic tractograms. The similarity matrix of the tractograms is subjected to a clustering algorithm and the results are mapped back on the cortical surface. In particular, we aim at effectively using the results of project 1 (Tractography and Local Modelling) and at developing new effective clustering algorithms, based on e.g. hierarchical clustering and fuzzy clustering. Furthermore, suitable measures for the quality of parcellations are being developed, which should reflect sharpness of the boundaries and internal homogeneity of patches. Our methods will be compared to other parcellation methods and applied to various areas of the cortex. [more]

The reconstruction of the fiber architecture in the brain's grey and white matter using diffusion MRI strongly depends on the quality of the images. Apart from the number of diffusion directions and the b-value, signal-to-noise ration and voxel resolution are the most important parameters. Ultra high field imaging using the MAGNETOM 7T scanner offers a possibility to improve these parameters. This, however, requires new acquisition techniques. In collaboration with the Neurophysics department (Dr. Robin Heidemann) such methods are being developed and tested. The obtained results form a quality in fiber imaging. [more]

Cortical and subcortical areas with different functions often differ with respect to their local tissue properties. This includes the arrangement and number of certain cell types (cytoarchitecture) and the course of local nerve fiber tracts (myeloarchitecture). These traits were already utilized by the early neuroanatomists, such as Brodmann, von Economo and Vogt, to chart the cortex. Diffusion weighted MRI allows for a non-invasive mapping of certain direction-dependent aspects of local microstructure (axons, dendrites). This way a subdivision (parcellation) of the cortex or subcortical structures can be achieved. We successfully applied this method to the basal ganglia, the amygdala and the thalamus, and showed in principle that it is also applicable to the cortex. [more]

In this research project we investigate quantitative aspects of diffusion based anatomical connectivity. On the one hand, we use Bayesian inference in order to estimate probability distributions for the number of fibers between two brain areas. On the other hand we develop techniques to map diffusion based local structural measures along reconstructed fiber tracts. [more]

This project is dedicated to the development of fast and versatile visualization of results of deterministic tractography, including effective selection of fibres, to effective visualization of probabilistic tractograms, including surface texturing, and to the integration of fibre visualization with the display of MRI, functional MRI and EEG/MEG source localization results. Moreover, we work on browser based visualization techniques that allow to convey information in a simple way by interactive 3d figures. [more]

In this research project, we deal with the evaluation of simple, yet biophysically realistic, models, so-called neural mass model, with respect to their dynamics. We investigate, if and under what circumstances certain phenomena that can be observed in EEG and MEG, such as oscillations or chaos, can be explained by these models, and how cognitive processes might be implemented. For the description of the dynamics we use bifurcation analysis, under consideration of input and synaptic parameters. Moreover, we investigate suitable ways to model transmission delays between neural populations, as well as the application of these models to EEG phenomena and human behavior. [more]

Neural field models constitute an extension of neural mass models by adding a spatial dimension. The allow for biologically realistic modelling of connectivity patterns. We aim at the application of these models for the explanation of phenomena in brain data (EEG and resting state fMRI) and behaviour. Moreover, we implement learning rules and investigate self organization of neural networks. [more]

This project deals with the incorporation of tissue inhomogeneity and anisotropy into physical head models in order to achieve more accurate solutions of the neuroelectromagnetic forward in inverse problems. The main goal is to investigate the impact of simplifications and model misspecifications onto the accuracy of forward calculation and source localization. The result of this project should be a comprehensive account on the sensitivity of forward calculations and source reconstructions towards possible inaccuracies in the forward model, depending on source position, noise, sensor configuration and other parameters. [more]