SummaryIn this study, we aim to cross-sectionally investigate lipid deposition and tissue morphology of the pancreas from magnetic resonance imaging (MRI) in people at risk of developing type 2 diabetes (T2D) compared to those without. To achieve this, we will deploy state-of-the-art artificial intelligence (AI)-based tools for automatic detection and segmentation of pancreas volume from magnetic resonance imaging (MRI) scans. Following this, we will extract imaging features and link them to the glycaemic status of individuals with normoglycaemia and prediabetes. Prospective analysis will be carried out to evaluate whether changes in pancreas tissue morphology and lipid deposition are related to glucose leveles across the population. The study will involve curation and analysis of the NAKO MRI database, leading to the identification of the imaging biomarkers associated with the plasma glucose levels obtained from the glucose tolerance test (OGTT) and glycated haemoglobin (HbA1c).

Keywords MRI deep-learning insulin-secretion pancreas prediabetes radiomics

InstitutionsLeibniz Center for Diabetes Research at the Heinrich Heine University Düsseldorf, Deutsches Forschungszentrum für Gesundheit und Umwelt (GmbH), Institut für Epidemiologie, Universitätsklinikum Heidelberg, University of Freiburg, University Hospital Heidelberg, Universitätsmedizin Greifswald, Section on Experimental Radiology, Deutsches Institut für Ernährungsforschung Potsdam-Rehbrücke, Deutsches Diabetes-Zentrum (DDZ), Leibniz-Zentrum für Diabetes-Forschung an der Heinrich-Heine-Universität Düsseldorf, Diagnostische und Interventionelle Radiologie, Deutsches Diabetes-Zentrum (DDZ), Universitätsklinikum Freiburg, Deutsches Diabetes Zentrum, Leibniz Center for Diabetes Research at the Heinrich Heine