Multimodal and Multiscale Deep Neural Networks for the Early Diagnosis of Alzheimer’s Disease using structural MR and FDG-PET images
Authors
Donghuan Lu, Simon Fraser University
Karteek Popuri, Simon Fraser University
Gavin Weiguang Ding, Simon Fraser University
Rakesh Balachandar, Simon Fraser University
Mirza Faisal Beg, Simon Fraser University
Michael Weiner, UCSF School of Medicine
Paul Aisen, UC San Diego School of Medicine
Ronald Petersen, Mayo Clinic
Cliford Jack, Mayo Clinic
William Jagust, University of California, Berkeley
John Trojanowki, University of Pennsylvania
Arthur Toga, University of Southern California
Laurel Beckett, University of California, Davis
Robert Green, Harvard Medical School
Andrew Saykin, Indiana University Bloomington
John C. Morris, Washington University in St. Louis
Leslie Shaw, Washington University in St. Louis
Jefrey Kaye, Oregon Health & Science University
Joseph Quinn, Oregon Health & Science University
Lisa Silbert, Oregon Health & Science University
Betty Lind, Oregon Health & Science University
Raina Carter, Oregon Health & Science University
Sara Dolen, Oregon Health & Science University
Lon Schneider, University of Southern California
Sonia Pawluczyk, University of Southern California
Mauricio Beccera, University of Southern California
Liberty Teodoro, University of Southern California
Bryan Spann, University of Southern California
James Brewer, University of California, San Diego
Helen Vanderswag, University of California, San Diego
Adam Fleisher, University of California, San Diego
Judith Heidebrink, University of Michigan, Ann Arbor
Publication Date
12-1-2018
Journal
Scientific Reports
URL with Digital Object Identifier
10.1038/s41598-018-22871-z
Abstract
Alzheimer’s Disease (AD) is a progressive neurodegenerative disease where biomarkers for disease based on pathophysiology may be able to provide objective measures for disease diagnosis and staging. Neuroimaging scans acquired from MRI and metabolism images obtained by FDG-PET provide in-vivo measurements of structure and function (glucose metabolism) in a living brain. It is hypothesized that combining multiple different image modalities providing complementary information could help improve early diagnosis of AD. In this paper, we propose a novel deep-learning-based framework to discriminate individuals with AD utilizing a multimodal and multiscale deep neural network. Our method delivers 82.4% accuracy in identifying the individuals with mild cognitive impairment (MCI) who will convert to AD at 3 years prior to conversion (86.4% combined accuracy for conversion within 1–3 years), a 94.23% sensitivity in classifying individuals with clinical diagnosis of probable AD, and a 86.3% specificity in classifying non-demented controls improving upon results in published literature.