New statistical methods for analysing resting state functional magnetic resonance imaging to detect localised differences

Functional magnetic resonance imaging (fMRI) is a widely used neuroimaging technique used to study brain activity through the blood oxygenation level-dependent (BOLD) signal in the brain. While task-based fMRI is more commonly used to examine neural response to specific stimuli, resting state fMRI (rs-fMRI) of spontaneous brain activity has been garnering more attention in recent years. More popular methods for rs-fMRI seek to identify functional connectivity between brain regions. Although effective at mapping global brain networks, connectivity-based approaches can overlook localised information contained in region-level signals. In this thesis, we explore alternative analyses that emphasize local and region-specific signal characteristics. We use various statistical measures to characterise the rs-fMRI signal and statistically test their ability to differentiate between patients with cognitive decline and control groups. Specifically, seven statistical measures are examined: mean...Show moreFunctional magnetic resonance imaging (fMRI) is a widely used neuroimaging technique used to study brain activity through the blood oxygenation level-dependent (BOLD) signal in the brain. While task-based fMRI is more commonly used to examine neural response to specific stimuli, resting state fMRI (rs-fMRI) of spontaneous brain activity has been garnering more attention in recent years. More popular methods for rs-fMRI seek to identify functional connectivity between brain regions. Although effective at mapping global brain networks, connectivity-based approaches can overlook localised information contained in region-level signals. In this thesis, we explore alternative analyses that emphasize local and region-specific signal characteristics. We use various statistical measures to characterise the rs-fMRI signal and statistically test their ability to differentiate between patients with cognitive decline and control groups. Specifically, seven statistical measures are examined: mean without cerebrospinal fluid regression, standard deviation (SD), amplitude of low frequency fluctuations (ALFF), fractional ALFF, spectral centroid, spectral variance, and spectral skewness. From our dataset, SD of the BOLD signal seems to be the most informative statistical measure in differentiating patients from controls. Moreover, the SD shows potential for tracking cognitive decline progress. While several other statistical measures show promise in differentiating specific cognitive decline diagnoses , they appear to be less suitable to track decline. These findings suggest that local and region-specific signal characteristics may contain relevant information on local brain function and dysfunction and can complement functional connectivity research to provide more insights into brain function and dysfunction.Show less

Leiden University Student Repository

Documents

In Collections

New statistical methods for analysing resting state functional magnetic resonance imaging to detect localised differences