Deep learning for neuroimaging: a validation study

doi:10.3389/fnins.2014.00229

Deep learning for neuroimaging: a validation study

2014-08-20

410

OA

AI

DOI：10.3389/fnins.2014.00229

Original

Save

View PDF

Abstract

En 中文

Deep learning methods have recently made notable advances in the tasks of classification and representation learning. These tasks are important for brain imaging and neuroscience discovery, making the methods attractive for porting to a neuroimager's toolbox. Success of these methodsis, in part, explained by the flexibility of deep learning models. However, this flexibility makes the process of porting to new areas a difficult parameter optimization problem. In this work we demonstrate our results (and feasible parameter ranges) napplication of deep learning methods to structuraland functional brain imaging data. These methods include deep belief networks and their building block the restricted Boltzmann machine. We also describe a novel constraint-based approachto visualizing high dimensional data. We use it to analyze the effect of parameter choices on data transformations. Our results show that deep learning methods are able to learn physiologically important representations and detect latent relations in neuroimaging data.

Keywords:

MRI

fMRI

intrinsic networks

classification

unsupervised learning

Deep learning for neuroimaging: a validation study

Abstract

AI Summary

Journal

Organization