Connectivity Based Parcellation and Brain Atlas Generation – Extracting Connectome Information for Schizophrenia Research

Title of thesis: CONNECTIVITY BASED PARCELLATION AND BRAIN ATLAS GENERATION – EXTRACTING CONNECTOME INFORMATION FOR SCHIZOPHRENIA RESEARCH Qi Wang, Master of Science, 2014 Thesis directed by: Professor Joseph JaJa Department of Electrical and Computer Engineering Traditional brain atlases are mainly based on hand-crafted anatomical structures, not taking into consideration useful connectivity pattern information. In our work, we use diffusion weighted imaging data to incorporate connectivity information into atlas generation. We use the software package FSL to process data to extract the connectivity matrix. The brain parcellation problem is formulated as a min-cut problem on a large, sparse graph. Spectral clustering and an original multi-class Hopfield network (MHN) method are applied to solve the problem, each working with a different analytical framework: MHN works in the diffusion space to generate individual parcellations, while spectral clustering works on standard space averaged connectome to generate group level atlases. Group study of brain images with schizophrenia is conducted, showing significant improvement in accuracy for disease diagnosis using features extracted with the proposed parcellation scheme. Hypothesis tests are performed on local structures to explore possible structural causes of the disease. CONNECTIVITY BASED PARCELLATION AND BRAIN ATLAS GENERATION – EXTRACTING CONNECTOME INFORMATION FOR SCHIZOPHRENIA RESEARCH