Files
Abstract
The study of magnetic resonance imaging (MRI) data hinges on a strong methodological foundation. With improvements in data collection, methodologies must evolve to keep pace with cutting-edge studies. One such area of study is in the graphical analysis of functional MRI data. This uses neuroactivational patterns inherent in data to determine which areas are most-highly linked with each other.To define these changes, this work begins by exploring ways to increase sensitivity of pre-existing methods. Using the difference degree test (DDT), it is shown that a modification to the null-model generation method results in higher sensitivity while maintaining accuracy of the findings. This results in the detection of differentially weighted edges which, in turn, help define differentially expressed hub nodes between groups.
Common brain parcellation strategies used in graphical studies often involve the application of an idealized brain parcellation which subdivides the brain into discreet regions based on anatomical landmarks. In practice, idealized anatomical parcellations are flawed due to the brain regions being based on idealized anatomical structures which are not representative of experimental data. A novel method referred to as the independent component analysis (ICA)-based parcellation algorithm (IPA) was developed to overcome these shortcomings. This method employed ICA to detect areas that exhibited similar neuroactivational patterns and subsequently used them to define regions of interest (ROIs). The IPA showed high consistency in ROI definition and showed higher homogeneity than the standard idealized anatomical parcellation and its usefulness in graph theoretic analysis.
After development using humans, the performance of an improved IPA known as the anatomy-free IPA (af-IPS) is profiled using a developmental pig traumatic brain injury model. The improvements made to the IPA produced regions of interest which maintained spatial consistency while also removing its dependence on idealized anatomy. Groupwise differences between treatment groups were profiled utilizing the subsequent parcellations.
Both the modified DDT and the IPA represent a substantive continuation of decades of research and methodological refinement in the field. While further research must be done to fully explore their potential, the DDT and IPA can hopefully serve as a springboard for further evolution in analytic methodologies.