Support vector classification analysis of resting state functional connectivity fMRI

Since its discovery in 1995 resting state functional connectivity derived from functional \nMRI data has become a popular neuroimaging method for study psychiatric disorders. \nCurrent methods for analyzing resting state functional connectivity in disease involve \nthousands of univariate tests, and the specification of regions of interests to employ in the \nanalysis. There are several drawbacks to these methods. First the mass univariate tests \nemployed are insensitive to the information present in distributed networks of functional \nconnectivity. Second, the null hypothesis testing employed to select functional connectivity \ndierences between groups does not evaluate the predictive power of identified functional \nconnectivities. Third, the specification of regions of interests is confounded by experimentor \nbias in terms of which regions should be modeled and experimental error in terms \nof the size and location of these regions of interests. The objective of this dissertation is \nto improve the methods for functional connectivity analysis using multivariate predictive \nmodeling, feature selection, and whole brain parcellation. \nA method of applying Support vector classification (SVC) to resting state functional \nconnectivity data was developed in the context of a neuroimaging study of depression. \nThe interpretability of the obtained classifier was optimized using feature selection techniques \nthat incorporate reliability information. The problem of selecting regions of interests \nfor whole brain functional connectivity analysis was addressed by clustering whole brain \nfunctional connectivity data to parcellate the brain into contiguous functionally homogenous \nregions. This newly developed famework was applied to derive a classifier capable of \ncorrectly seperating the functional connectivity patterns of patients with depression from \nthose of healthy controls 90% of the time. The features most relevant to the obtain classifier \nmatch those previously identified in previous studies, but also include several regions not \npreviously implicated in the functional networks underlying depression.