Data driven extraction of complementary structural MRI biomarkers for
predicting conversion to dementia 
   Lead Investigator:    Dante Alberto De Nigris Moreno
   Institution      :    Arctic Fox AI
   E-Mail           :    dante@arcticfox.ai
   Proposal ID      :    1058



   Proposal Description:
   The aim of this project is to implement and evaluate data driven machine learning and deep learning methods which extract complementary biomarkers from structural MRI (e.g. texture descriptor of the hippocampus) with the interest of improving the accuracy of predicting whether a presymptomatic patient converts to dementia. We will define as baseline performance the prediction accuracy obtained exclusively with the imaging biomarkers readily available in the Imaging Data (e.g. volumetrics). We will also evaluate the improvement in prediction accuracy by additionally incorporating CSF and APOE polymorphism as additional input features to the final classification model. Finally, we are also requesting pathology data to determine the accuracy of structural MRI biomarkers to predict final pathological status, as opposed to a future clinical diagnosis.