The Brain and Data Science Group

Head: Dr. rer. nat. Stefan Haufe

Our group works on the intersection of data science and neuroimaging. We develop algorithms to analyze EEG, MEG, and other electrophysiological data. The main advantage of these recordings is their high temporal resolution, which enables us to study dynamical properties of neuronal populations inaccessible to fMRI. Moreover, being non-invasive, they are of interest for diagnosing neurological and psychiatric disorders.

The main disadvantage of EEG and MEG is that both capture mixtures of many different brain processes at the sensor level. This can lead to serious misinterpretations about the locations and the functional connectivity profile of the underlying brain sources. Our group develops advanced signal processing and unsupervised machine learning techniques to reconstruction the underlying source activations and their connectivity profile as accurately as possible, and thereby to enable correct neurophysiological interpretation.  We explore this possibility by applying our methods to clinical data in collaboration with leading medical experts. We are currently interested in movement disorders, dementias, developmental disorders, and aging. The goal of this research is to derive neural biomarkers of mental disorders. To achieve that, we leverage powerful machine learning algorithms to derive predictions of clinical variables from multivariate electrical source imaging and functional connectome data. A particular focus of our machine learning work is the correct neurophysiological interpretation of predictive models.

Lab Setting

Our group is located at the Berlin Center for Advanced Neuroimaging (BCAN), directed by Prof. John-Dylan Haynes, on the historic Charité Campus Mitte (CCM) in the center of Berlin. We are situated right above two Siemens 3T MRI scanners equipped with concurrent EEG setup. Our CPU and GPU servers allow us to run complex analyses on big datasets in little time.

We are associated with the Bernstein Center for Computational Neuroscience (BCCN), and we closely collaborate with the Movement Disorders Group of Prof. Andrea Kühn at Charité Neurology, the Neural Interactions and Dynamics group of Prof. Vadim Nikulin at MPI for Cognitive and Brain Sciences, Leipzig, the Magnetoencephalography Group of Dr. Guido Nolte at UKE Hamburg, the Methods of Plasticity Research Group of Prof. Nicolas Langer at University of Zurich, and the Machine Learning Group of Prof. Klaus-Robert Müller at TU Berlin, among others. We are funded by an ERC starting grant of the European Union.