Support vector classification analysis of resting state functional connectivity fMRI

Machine learning;Support vector classification;FMRI;Biological signal processing

Craddock, Richard Cameron ; Electrical and Computer Engineering

University:Georgia Institute of Technology

Department:Electrical and Computer Engineering

关键词: Machine learning;    Support vector classification;    FMRI;    Biological signal processing;   

Others  :  https://smartech.gatech.edu/bitstream/1853/31774/1/craddock_richard_c_200912_phd.pdf

美国|英语

来源: SMARTech Repository

【 摘 要 】

Since its discovery in 1995 resting state functional connectivity derived from functionalMRI data has become a popular neuroimaging method for study psychiatric disorders.Current methods for analyzing resting state functional connectivity in disease involvethousands of univariate tests, and the specification of regions of interests to employ in theanalysis. There are several drawbacks to these methods. First the mass univariate testsemployed are insensitive to the information present in distributed networks of functionalconnectivity. Second, the null hypothesis testing employed to select functional connectivitydierences between groups does not evaluate the predictive power of identified functionalconnectivities. Third, the specification of regions of interests is confounded by experimentorbias in terms of which regions should be modeled and experimental error in termsof the size and location of these regions of interests. The objective of this dissertation isto improve the methods for functional connectivity analysis using multivariate predictivemodeling, feature selection, and whole brain parcellation.A method of applying Support vector classification (SVC) to resting state functionalconnectivity data was developed in the context of a neuroimaging study of depression.The interpretability of the obtained classifier was optimized using feature selection techniquesthat incorporate reliability information. The problem of selecting regions of interestsfor whole brain functional connectivity analysis was addressed by clustering whole brainfunctional connectivity data to parcellate the brain into contiguous functionally homogenousregions. This newly developed famework was applied to derive a classifier capable ofcorrectly seperating the functional connectivity patterns of patients with depression fromthose of healthy controls 90% of the time. The features most relevant to the obtain classifiermatch those previously identified in previous studies, but also include several regions notpreviously implicated in the functional networks underlying depression.