Resting State fMRI Functional Connectivity-Based Classification Using a Convolutional Neural Network Architecture

• Published in: Frontiers in neuroinformatics Vol. 11; p. 61
• Main Authors: Meszlényi, Regina J., Buza, Krisztian, Vidnyánszky, Zoltán
• Format: Journal Article
• Language: English
• Published: Switzerland Frontiers Research Foundation 17.10.2017; Frontiers Media S.A
• Subjects: Artificial intelligence; Brain; Classification; Cognitive ability; connectome; convolutional neural network; Datasets; Deep learning; Design; Dynamic Time Warping; Functional magnetic resonance imaging; Learning algorithms; Machine learning; Neighborhoods; Neural networks; Neuroscience; resting state connectivity; Sparsity; Hungary; connectome; Dynamic Time Warping; classification; resting state connectivity; convolutional neural network; functional magnetic resonance imaging
• DOI: 10.3389/fninf.2017.00061
• ISSN: 1662-5196

Machine learning techniques have become increasingly popular in the field of resting state fMRI (functional magnetic resonance imaging) network based classification. However, the application of convolutional networks has been proposed only very recently and has remained largely unexplored. In this paper we describe a convolutional neural network architecture for functional connectome classification called connectome-convolutional neural network (CCNN). Our results on simulated datasets and a publicly available dataset for amnestic mild cognitive impairment classification demonstrate that our CCNN model can efficiently distinguish between subject groups. We also show that the connectome-convolutional network is capable to combine information from diverse functional connectivity metrics and that models using a combination of different connectivity descriptors are able to outperform classifiers using only one metric. From this flexibility follows that our proposed CCNN model can be easily adapted to a wide range of connectome based classification or regression tasks, by varying which connectivity descriptor combinations are used to train the network.