Utilizando Aprendizado de Máquina para Predição de Diagnósticos da Doença de Alzheimer

Abstract

Alzheimer’s disease is a gradual and irreversible neurodegenerative disease with an insidious onset that causes memory loss and a variety of cognitive problems. Currently, it is the most common type of dementia in the world, representing about 60% to 80% of cases. There is no simple way to detect it. Diagnosis requires a complete medical examination. Early diagnosis gives the patient and family a better chance of slowing the progression of the disease and having more control over the symptoms. While there is no cure or way to stop its progression and treatment is still an open research question, there are pharmacological and non-pharmacological options that can help treat the symptoms. Early detection through conventional magnetic resonance imaging will facilitate timely and effective interventions/treatments, which would increase patients’ life expectancy and quality of life. Machine learning is an area of computer science research that involves the induction of hypotheses or functions, from past experiences, to solve a given problem of a predictive or descriptive nature. With the advancement of machine learning technology development, it can be used as a quick auxiliary diagnostic method. Deep learning is a sub-area of machine learning, it can be thought of as an automated form of predictive analysis, which emphasizes successive layers with increasingly meaningful representations. This work proposes to assist in the accurate diagnosis of Alzheimer’s disease using machine learning and deep learning. The main objective of this work is to propose and evaluate a solution for the detection of Alzheimer’s disease diagnoses in magnetic resonance images, using machine learning and deep learning techniques. In this way, 15 machine learning models and 4 different convolutional neural network architectures were implemented.