Statistical disease detection with resting state functional magnetic resonance imaging

Öztürk, Ebru
Most of the functional magnetic resonance imaging (fMRI) data are based on a particular task. The fMRI data are obtained while the subject performs a task. Yet, it's obvious that the brain is active even when the subject is not performing a task. Resting state fMRI (R-fMRI) is a comparatively new and popular technique for assessing regional interactions when a subject is not performing a task. This study focuses on classifying subjects as healthy or diseased with the diagnosis of schizophrenia by analyzing R-fMRI data. The resting state situation in the dataset of “UCLA Consortium for Neuropsychiatric Phonemics LA5c Study” is used to extract brain signals in the Region of Interest (ROI) analysis. The default mode network (DMN) ROIs were selected since the DMN is a perception depending on an interconnected set of areas displaying higher activity during rest than task related activity (Raichle and Snyder, 2007). Pre-processing of fMRI images is achieved with toolbox of Statistical Parametric Mapping version 8 (SPM8). ROI-based on brain signals are obtained from Functional Connectivity (CONN). After brain signals are obtained, the disease status is predicted by adjusting for the magnitude of brain signals, the demographic information’s of subjects such as gender and age. Logistic regression model, marginal model, random effect model and k-means clustering, hierarchical clustering and clustering genes with replications (CGR) followed by logistic regression approaches are conducted to classify the subjects in the UCLA data set by using R-Studio. Marginal model with smoking status and k-means clustering algorithm followed with logistic regression model excluding smoking status give best results.