S-123 | Enhancing Alzheimer’s Diagnosis through fMRI Analysis using Variational Autoencoders

Theoretical and Computational Neuroscience
Author: Santiago Valentino Blas Laguzza | Email: santiblaas@gmail.com

Santiago Valentino Blas Laguzza^1°, Martin Alberto Belzunce^2°, Diego Diego^1°

^1° Intituto de matematica aplicada del litoral IMAL-CONICET
^2° ICIFI UNSAM-CONICET, Escuela de Ciencia y Tecnología, UNSAM

My research focuses on improving the diagnosis of Alzheimer’s disease by analyzing fMRI data using Variational Autoencoders (VAEs). This approach leverages the VAEs’ ability to capture complex patterns in high-dimensional data by mapping it into a latent space. The latent space representation allows for the differentiation between Alzheimer’s patients and healthy individuals. By analyzing the properties of this latent space, we can better understand the underlying neural changes associated with Alzheimer’s, which may serve as biomarkers for early detection. Additionally, I explore the use of graph-based methods to interpret these biomarkers more effectively. This combination of advanced machine learning techniques and neuroscience aims to enhance diagnostic accuracy and provide deeper insights into the disease’s progression. My work also involves investigating the application of transformers and LSTM networks in processing temporal sequences of clinical data, further pushing the boundaries of computational neuroscience.