Premium
P2‐073: FEASIBILITY OF QUANTIFYING AMYLOID BURDEN USING VOLUMETRIC MRI DATA: PRELIMINARY FINDINGS BASED ON THE DEEP LEARNING 3D CONVOLUTIONAL NEURAL NETWORK APPROACH
Author(s) -
Yuan Ye,
Wang Zhangyang,
Lee Wendy,
Thiyyagura Pradeep,
Reiman Eric M.,
Chen Kewei
Publication year - 2018
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1016/j.jalz.2018.06.758
Subject(s) - convolutional neural network , pattern recognition (psychology) , artificial intelligence , deep learning , pooling , mean squared error , artificial neural network , neuroimaging , computer science , positron emission tomography , mathematics , nuclear medicine , statistics , medicine , psychology , neuroscience
study, we found the cutoff point for amyloid PET (PIB) and fluorodeoxyglucose (FDG) PET is an SUVR of 1.1 (SD 0.17, 95% CI [0.9, 1.5]) and 1.7 (SD 0.1, 95% CI [1.6, 2.0]) respectively. Similar analysis will be conducted for other biomarkers. Conclusions:The method provides a straightforward statistical approach to determine the biomarker cutoff points independent of participants’ disease status and amyloid status. It can be applied to any biomarkers in both sporadic AD and DIAD, and can easily account for covariates.