Simone Bonechi (DIISM, University of Siena)

Mar 6, 2019 – 11:00 AM

DIISM, Artificial Intelligence laboratory (room 201), Siena SI

Description

In recent years, deep learning and Convolutional Neural Networks (CNNs) have produced a devastating impact on computer vision,
achieving outstanding results on a variety of problems, including medical image analysis. Recently, these techniques have been extended
to 3D images with the downside of an high increase of the computational burden. This kind of convolutions are used in most of the
state–of–the–art CNNs for the analysis of human Magnetic Resonance Imaging (MRI) of the brain. One important task in the analysis of MRI
brain images is the age prediction, since it has been demonstrated that a discrepancy between the estimated age of the brain and the
chronological age is correlated with cognitive disease like some kind of dementia and Alzheimer. In this paper, we propose an effective
alternative to 3D convolution that allows to reduce the computational requirements for this kind of analysis. The proposed network allows to obtain comparable results with other concurrent methods requiring only a fraction of training time and GPU memory.