Machine-learning guided radiological feature extraction to differentiate LATE
with idiopathic Alzheimer's Disease 
   Lead Investigator:    Bryce Tan
   Institution      :    National University of Singapore
   E-Mail           :    bryce_tan@nuhs.edu.sg
   Proposal ID      :    1294



   Proposal Description:
   Limbic-predominant age-related TDP-43 encephalopathy (LATE) is an emerging clinical  phenotype that has a unique pathophysiological process compared to idiopathic Alzheimer's  Disease in which amyloid burden plays a major role in neurodegeneration. While there are  emerging radiological features of LATE differentiating itself from Alzheimer's Disease, it remains  challenging to differentiate radiologically between subjects with AD and LATE. We predict that  machine-learning guided radiological feature extraction could provide a novel approach to  extract key radiomic features that could accurately differentiate LATE from AD, and to validate  in known clinical datasets with available biomarker data to investigate potential biomarkers  that can accurately differentiate the two different, yet clinically similar, syndromes.    To do this, we propose the following:  - We will subject MRI images to radiomic feature extraction, and subject them to machinelearning algorithms. This will be subjected to validation in other large datasets, and will be  compared with histopathological and biochemical biomarkers. Algorithms developed during the  course of the study will be used to enhance clinical diagnostics to examine this emerging clinical  phenotype.