High-throughput preprocessing pipeline for resting-state fMRI

Abstract

Resting-state functional magnetic resonance imaging (fMRI) examines functional connectivity between brain regions by measuring spontaneous fluctuation of blood-oxygen-level dependent (BOLD) signal while subjects are at rest. Recent studies have revealed that neurodegenerative diseases are associated with abnormalities in resting-state functional connectivity. This thesis work aimed to develop a high-throughput preprocessing pipeline for minimizing spatial and temporal artifacts in resting-state fMRI data. The building blocks and cluster computing capabilities of the pipeline are discussed in detail. In an effort to validate this pipeline, a seed-based analysis was performed on preprocessed data using 3 seeds placed in the posterior cingulate cortex (PCC), supplementary motor area (SMA) and inferior parietal sulcus (IPS). These 3 seeds represent the core components of the default mode network (DMN) and task-positive network (TPN), two of the most commonly examined resting-state networks. Consistent with literature, results indicate that activity in the PCC is functionally correlated to regions of the DMN and anti-correlated to regions of the TPN, whereas activities in the SMA and IPS are functionally correlated to regions of the TPN and anti-correlated to regions of the DMN. These functional connectivity patterns were consistent across scans on both group-level and subject-level. This preprocessing pipeline will support researchers and clinical collaborators of Medical Image Analysis Lab (MIAL) in their analysis of resting-state functional connectivity.