TrueBrainConnect: predicting brain disorders
New ERC-funded project at Charité to combine EEG data and machine learning
Hosted by Charité – Universitätsmedizin Berlin, the TrueBrainConnect project aims to systematically study connections between different areas of the brain, and hopes to draw conclusions regarding potential disease patterns. A team of researchers, led by Dr. Stefan Haufe – a computer scientist and expert in the field of machine learning – are hoping to produce complex models capable of predicting different mental states. Their research will be based on EEG data. The new working group has secured funding from the European Research Council (ERC) for a duration of five years, and will be based at Charité’s Berlin Center for Advanced Neuroimaging.
The TrueBrainConnect project is based on the premise that many neurological disorders start to manifest themselves even before the onset of symptoms, or before they produce noticeable changes in brain structure or behavior. The project’s hypothesis is that these disorders herald their presence through irregularities in the way different areas of the brain communicate with each other. The new working group, which will be led by Dr. Haufe, will try to develop methods capable of reliably estimating and localizing brain interactions, in the hope of improving the prognosis of pathological conditions affecting the brain. “Electroencephalography (EEG), which records the brain’s electrical activity, is one of the most well-established neuroimaging methods, and allows researchers to study brain activity almost in real time. The data generated by this process are entered into computer models, which allow us to deduce the principles guiding these interactions within the brain,” explains Dr. Haufe.
The methods currently available for the analysis of neuroimaging data are not yet sufficiently developed, and robust conclusions cannot be drawn from them. It is hoped that new signal processing and machine learning techniques will allow researchers to make precise determinations regarding brain signal sources and actual nerve cell interactions. To this end, Dr. Haufe and his team will be working in close collaboration with clinical colleagues: “The new TrueBrainConnect method will be capable of visualizing neuronal patterns that underpin disorders such as dementia and Parkinson’s disease. The primary aims of these endeavors include the identification of the early precursors of disease and the delineation of different types of dementia.” The researchers are hopeful that this will help improve the early diagnosis of these extremely common neurological disorders.
However, numerous challenges will need to be overcome along the way. For instance, for each signal recorded, researchers must determine whether its origins can be clearly traced back to an actual interaction between different areas of the brain, or whether it has a different origin. While EEG offers researchers numerous advantages, the technology also presents technical challenges that need to be resolved, such as poor spatial resolution and a high degree of noise contamination. “To improve the quality of predictions, TrueBrainConnect will use the latest methods from the field of machine learning. One important question the project will be investigating is how to correctly interpret these new models; how to recognize which patterns of brain activity are crucial for a particular prognosis,” explains Dr. Haufe.
“Dr. Haufe’s research fills an important gap, and places EEG research on methodologically sound foundations,” says Prof. Dr. John-Dylan Haynes, Director of the Berlin Center for Advanced Neuroimaging and researcher at Charité’s Bernstein Center for Computational Neuroscience. TrueBrainConnect will be positioned at the interface between two high-profile fields of research: the analysis of functional brain connectivity and the study of age-related neurological disorders. Both the development of improved statistical techniques for the analysis of neuroimaging data, as well as the development of new methods of signal processing and machine learning, will form the heart of these endeavors. Their practical application, and the evaluation of their effectiveness in the investigation of neurological disorders, will form part of the second stage of this project. It is anticipated that these new techniques will be used in the fields of Brain-Computer Interfacing and Mental State Monitoring, as well as in basic and clinical research.
Aimed at early-career scientists, ERC Starting Grants are awarded by the European Research Council as part of the Horizon 2020 Framework Programme for Research and Innovation. A total of €1.5 million has been secured to fund the development of the working group at Charité’s Berlin Center for Advance Neuroimaging (Grant Agreement n°758985).
Deputy Press Officer
Charité – Universitätsmedizin Berlin
Tel: +49 30 450 570 400
Berlin Center for Advanced Neuroimaging