Representation of human brain by mesh networks

Önal Ertuğrul, Itır
In this thesis, we propose novel representations to extract discriminative information in functional Magnetic Resonance Imaging (fMRI) data for cognitive state and gender classification. First, we model the local relationship among a set of fMRI time series within a neighborhood by considering temporal information obtained from all measurements in time series. The estimated local relationships, called Mesh Arc Descriptors (MADs), are employed to represent information in fMRI data. Second, we adapt encoding methods frequently used in Computer Vision, namely Fisher Vectors (FV), Vector of Locally Aggregated Descriptors (VLAD) and Bag-of-Words (BoW) to encode local MADs. We show that employing MADs outperform state-of-the-art fMRI representations and encoding them further with FV gives superior performance over MADs. Then, we propose a hierarchical framework, called Hierarchical Multi-resoution Mesh Networks (HMMNs), in which the fMRI signal is decomposed into multiple subbands and mesh networks are constructed for each subband separately. We fuse the decisions of classifiers trained with multi-resolution mesh-networks in the final step of the framework. We show that Hierarchical Multi-resolution Mesh Networks outperform mesh-networks constructed from original fMRI signal. Finally, we adapt multi-resolution approach for gender classification using fMRI data. We fuse the decisions of classifiers trained with multi-resolution multi-task mesh networks in a 2-level hierarchical architecture to discriminate gender. The proposed gender classification framework performs better compared to single layer architectures fusing only multi-task or only multi-resolution mesh networks.  
Citation Formats
I. Önal Ertuğrul, “Representation of human brain by mesh networks,” Ph.D. - Doctoral Program, Middle East Technical University, 2017.