Using Transfer Learning to classify different stages of Alzheimer’s disease

Abstract: The identification of Alzheimer’s disease through the application of various machine learning techniques on neuroimaging data of the likes of MRI is an area of study which has seen intense levels of research in recent years. While many machine learning techniques have existed for a long time, recent advances in Deep Learning and Computer Vision have allowed for better performing predictive models. In this report we present a Convolutional Neural Network & Transfer Learning based approach to classify various stages of Alzheimer’s disease with a high degree of accuracy. Furthermore we investigate how accurately Transfer Learning can transfer knowledge between more and less severe stages of Alzheimer’s disease. With the use of data augmentation techniques our Convolutional Neural Network approach managed to reach a predictive accuracy of 97%, while our Transfer Learning based approach managed to reach a predictive accuracy of 70%. Our research shows that with our approach we can reach comparable or even better performance than state- of- the-art Deep Learning based methods, despite using a significantly smaller dataset.