Brain MRI Image Analysis Using Deep Learning and Its Application to Diagnosis Support of Alzheimerʼs Disease

Abstract

It is known that the human brain atrophies with normal aging and it is possible to estimate the age from T1-weighted images by utilizing this fact. Brain disorders such as Alzheimerʼs disease produce a different pattern of atrophy from normal aging. We can support the diagnosis by comparing the age estimated from T1-weighed images with the actual age. In this paper, we propose a method for age estimation using 3D CNN. In the proposed method, we use 3D CNN to extract features suitable for age estimation and input the gender as supplementary information for a fully- connected layer to improve the estimation accuracy. We demonstrate the effectiveness of the proposed method through experiments using a T1-weighted image database consisting of healthy Japanese subjects. We also discuss the application of the proposed method to support the diagnosis of Alzheimerʼs disease.