A study of the biological sex in the classification of Alzheimer’s disease using a convolutional neural network

Abstract: This report investigates whether Alzheimer’s MRI scans could be classified more accurately using deep learning if the biological sex was considered. The data used in the study were female and male Alzheimer’s Disease (AD) and Cognitive Normal (CN) MRI scans. The data was divided into a training and test set. Three convolutional neural networks, based on the same architecture, were trained on different data, one on the female training data (female model), one on the male training data (male model), and one on both the female and male training data (combined model). The female model was tested on the female test data, the male model on the male test data and the combined model on first the female and then the male test data. The female model classified female AD and female CN pictures significantly more accurately than the combined model. The male model did not classify male AD and male CN pictures significantly more accurately. Since the female model achieved a significant difference, our study implies that there could be a difference in the brains of males and females regarding Alzheimer’s.