CMStatistics 2019: Start Registration
View Submission - CMStatistics
B2015
Title: Identifying aberrant EEG functional connectivity in schizophrenia using an ensemble of convolutional neural networks Authors:  Chee Ming Ting - King Abdullah University of Science and Technology (Saudi Arabia) [presenting]
Hernando Ombao - King Abdullah University of Science and Technology (KAUST) (Saudi Arabia)
Abstract: The aim is to leverage on altered patterns in brain functional connectivity as features for automatic discriminative analysis of neuropsychiatric patients. Deep learning methods have been introduced recently to fMRI functional network classification, however, existing architectures focused on a single type of connectivity measure. We propose a deep convolutional neural network (CNN) for classifying electroencephalogram (EEG)-derived brain connectome in schizophrenia (SZ). To capture complementary aspects of disrupted connectivity in SZ, we explore fusion of heterogeuous connectivity features consisting of time and frequency-domain metrics of effective connectivity based on vector autoregressive model and partial directed coherence, and complex network measures of network topology. We design a novel multi-domain connectome CNN (MDC-CNN) based on a parallel ensemble of 1D and 2D CNNs to integrate these features from various domains and dimensions using different fusion strategies. We also consider an extension to dynamic brain connectivity using the recurrent neural networks. Hierarchical latent representations learned by the multiple convolutional layers from EEG connectivity reveals apparent group differences between SZ and healthy controls (HC). Evaluated on resting-state EEG data, the proposed MDC-CNN by integrating information from diverse brain connectivity descriptors is able to accurately discriminate SZ from HC, outperforming support vector machines.