The “Connectivity“ group at the Institute of Neuroscience and Medicine (INM-1) of the Research Centre Jülich is interested in the structural and functional connectivity of the human brain in relation to its underlying structure. A particular focus lies on their changes during the aging process in relation to environmental and genetic influences. At this, we use data from large epidemiological population studies with thousands of subjects, which the group is leading or contributing to.
What is the data science project’s research question? Can cognitive performance be predicted from functional connectivity metrics using Deep Learning? Can prediction performance be improved if using not only derived metrics, but full raw brain data series (images or signals) as input for the Deep Learning framework?
What data will be worked on? Brain imaging data acquired with magnetic resonance imaging techniques (volume data and time series), derived functional connectivity metrics (based on graph theory), and cognitive performance measures.
What tasks will this project involve? Application of machine/deep learning methods on neuroimaging and related phenotypic data to detect aging-related patterns and achieve group classifications, including code development, data analyses, and joint publication.
What makes this project interesting to work on? The brain undergoes a variety of changes during aging accompanied by a decline in cognitive functions. Predicting this decline in cognitive performance from brain data is of high interest to come to a more mechanistic understanding of these age-associated changes. Subsequently, cross-validating this knowledge from basic aging research to age-associated disease-related processes in the brain is of high societal importance to improve healthy aging. First approaches to predict cognitive performance, also from our own group, showed that metrics derived from functional brain imaging data could generally be used to predict and classify differences in cognitive performance using classical machine learning algorithms but are rather limited in prediction performance and accuracy. Avoiding such data reduction is thus of utmost importance to leverage the full potential of this prediction task, i.e. using more comprehensive data features in a sophisticated machine or deep learning framework.
What is the expected outcome? Contribution to research paper
What infrastructure, programs and tools will be used? Can they be used remotely? Computing resources on high-performance computing cluster of the Jülich Supercomputing Center (JURECA), self-developed python code using TensorFlow, PyTorch, and/or Scikit-learn. All can be used remotely.
What skills are necessary for this project? Data analytics / statistics, Scientific computation, Data mining / Machine learning, Deep learning, High-performance computing, Knowledge of neuroimaging data processing and neuropsychological assessments are useful, but not necessary.
Is the data open source? No
Interested candidates should be at Postdoc-level . Svenja Caspers is looking for 2 visiting scientists, working on the project together with the team.