Automatic classification of AD using deep learning

 
   Lead Investigator:    Deqiang Qiu
   Institution      :    Emory University
   E-Mail           :    dqiu3@emory.edu
   Proposal ID      :    784



   Proposal Description:
   The goal of this project is develop a deep learning based algorithm for           classifying AD patients and other dementia. A deep learning neural network will   be trained using the MRI and other data from NACC as well as other sources.       Successful execution of the project will result in an objective classifier for    differentiating different types and levels of dementia, and thereby providing     insights into the disease processes.                                              Dr. Qiu is trained as an MRI physicist and neuroscientist, and have more 13       years of experience in the field. Dr. Qiu has the expertise in performing MRI     image analysis himself and in his lab.                                            Specific aim 1: We still evaluate differences in structural and physiological     measures derived from MRI, such as cortical thickness, volume of white matter     hyper intensities between different types of dementia and from normal controls.   Specific aim 2: We will correlate these MRI measures with other data such as      Abeta42 from CSF analysis, and genetic results. Specific Aim 3: We will train a   deep learning neural network using MRI derived measures and other measures. We    hypothesize that a higher sensitivity and specificity can be achieved by adding   MRI data.