Title: "Deep neural networks for analysis of fMRI data" 

In this talk, I will introduce the works in my lab on fMRI data analysis using neural networks-based machine learning models (Kim et al., 2016; Jang et al., 2017; Kim et al., 2019). Using the weight sparsity controlled deep neural networks (DNN), a method to classify a schizophrenic from healthy control has been proposed using whole-brain functional connectivity (FC) patterns (Kim et al., 2016). Then, a proof-of-concept for whole-brain voxel-wise pattern classification was introduced using four sensorimotor dataset (Jang et al., 2017). More recently, the DNN model has been applied to predict emotional scores of participants using whole-brain voxel-wise patterns evoked from auditory sounds (Kim et al., 2019). The proposed DNN model has been further updated to enhance a computational efficiency while maintaining performance and an alternative convolutional neural networks-based model has been investigated. If time allows, other research studies in my lab will be introduced such as denoising of EEG data simultaneously acquired with fMRI data, real-time fMRI-based neurofeedback, and a novel naturalistic viewing paradigm.

References

[Kim et al., 2016] Kim J, Calhoun VD, Shim E, Lee JH, Deep neural network with weight sparsity control and pre-training extracts hierarchical features and enhances classification performance: Evidence from whole-brain resting-state functional connectivity patterns of schizophrenia, Neuroimage. 2016 Jan 1;124(Pt A):127-46.

[Jang et al., 2017] Jang H, Plis SM, Calhoun VD, Lee JH. Task-specific feature extraction and classification of fMRI volumes using a deep neural network initialized with a deep belief network: Evaluation using sensorimotor tasks. Neuroimage. 2017 Jan 15;145(Pt B):314-328.

[Kim et al., 2019] Kim HC, Bandettini P, Lee JH, Deep neural network predicts emotional responses of the human brain from functional magnetic resonance imaging, Neuroimage. 2019 Feb 1;186:607-627.