Automated classification of MRIs into Parkinsons and Dementia related diseases

 
   Lead Investigator:    Shubham Jain
   Institution      :    Qure.ai
   E-Mail           :    shubham.jain@qure.ai
   Proposal ID      :    839



   Proposal Description:
   We want to use deep learning and classify MRIs into Parkinsons, AD, MCI and       Normal. Our current results with ADNI show 80+ accuracy on AD vs NL which we      want to use for this dataset.                                                     The aims of the research we are doing is to completely automate the               classification of the structural MRIs into normal and other neurodegenerative     diseases, specifically Alzheimer?s, MCI and Parkinsons. The main variables of     our interest are the diagnosis of the patient, age, MMSE, gender, ABETA, APOE     genotype if available and of course the MRI Image. We will be employing deep      learning techniques for classification and using FSL and ANTs for preprocessing   of the data. Main outcomes of interest would be how our classification works      with sex, age, MMSE and other provided variables.