RT Journal Article
T1 Studying the Manifold Structure of Alzheimer’s Disease: A Deep Learning Approach Using Convolutional Autoencoders
A1 Martínez-Murcia, Francisco Jesús
A1 Ortiz-García, Andrés
A1 Górriz-Sáez, Juan Manuel
A1 Ramírez, Javier
A1 Castillo-Barnes, Diego
K1 Alzheimer, Enfermedad de
AB Many classical machine learning techniques havebeen used to explore Alzheimer’s Disease, evolving from imagedecomposition techniques such as Principal Component Analysistowards higher-complexity, non-linear decomposition algorithms.With the arrival of the deep learning paradigm, it has becomepossible to extract high-level abstract features directly fromMRI images that internally describe the distribution of datain low-dimensional manifolds. In this work, we try a newexploratory data analysis of Alzheimer’s Disease (AD) basedon deep convolutional autoencoders. We aim at finding linksbetween cognitive symptoms and the underlying neurodegenera-tion process by fusing the information of neuropsychological testoutcomes, diagnoses and other clinical data with the imagingfeatures extracted solely via a data-driven decomposition ofMRI. The distribution of the extracted features in differentcombinations is then analysed and visualized using regressionand classification analysis, and the influence of each coordinateof the autoencoder manifold over the brain is estimated. Theimaging-derived markers could then predict clinical variableswith correlations above 0.6 in the case of neuropsychologicalevaluation variables such as the MMSE or the ADAS11 scores,achieving a classification accuracy over 80% for the diagnosis ofAD.
PB IEEE
YR 2020
FD 2020-01-01
LK https://hdl.handle.net/10630/28806
UL https://hdl.handle.net/10630/28806
LA eng
NO Martínez-Murcia, Francisco J. et al. “Studying the Manifold Structure of Alzheimer's Disease: A Deep Learning Approach Using Convolutional Autoencoders.” IEEE Journal of Biomedical and Health Informatics 24 (2020): 17-26.
DS RIUMA. Repositorio Institucional de la Universidad de Málaga
RD 19 ene 2026