Deep learning approach to predict conversion from mild cognitive impairment neuropsychological subtypes to Alzheimer's disease using MRI

• Grant number: 23/00327-1
• Support Opportunities: Scholarships abroad - Research Internship - Post-doctor
• Start date: June 01, 2023
• End date: November 30, 2023
• Field of knowledge: Health Sciences - Medicine
• Principal Investigator: Carlos Ernesto Garrido Salmon
• Grantee: Rodolfo Dias Chiari Correia
• Supervisor: James Cole
• Host Institution: Faculdade de Filosofia, Ciências e Letras de Ribeirão Preto (FFCLRP). Universidade de São Paulo (USP). Ribeirão Preto , SP, Brazil
• Institution abroad: University College London (UCL), England
• Associated to the scholarship: 21/14873-2 - Identification of mild cognitive impairment subtypes in the elderly using multimodality magnetic resonance imaging and unsupervised machine learning, BP.PD

Abstract

Artificial intelligence applications in medical images have grown significantly in recent years, with deep learning (DL) techniques being one of the most used and powerful techniques to identify patterns in this type of image. In the context of Alzheimer's disease (AD), there are two common approaches using DL and Magnetic Resonance Imaging (MRI). The first one is for disease detection purposes and the other one, which is the focus of our study, is to predict the conversion from mild cognitive impairment (MCI) to AD. However, the MCI stage in the elderly is very heterogeny and has unsatisfactory diagnostic accuracy, which makes this second approach complex and with inconsistent results. A strategy to reduce the heterogeneity of the MCI group is to subdivide it into subgroups and study them separately, being the subtypes based on neuropsychological characteristics one of the best established in the literature. The objective of this project is to use DL and MRI techniques to train a model to predict the conversion of the different neuropsychological MCI subtypes to AD. For this, we will use the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset from where we will extract demographic, clinical, neuropsychological, and MR imaging data. With the ADNI longitudinal data, we will use patients diagnosed with MCI at baseline separating them by neuropsychological subgroup. For each subgroup, we will train a model considering two classes, those that converted to AD and those that did not convert in the follow-up. As input data, we will use T1-weighted MR images of the medial regions of the temporal lobe. The idea is to see if this approach increases the ability to predict which individuals with MCI are more likely to convert to AD. (AU)