Functional brain networks exhibit complex spatiotemporal dynamics that are increasingly recognized as critical to understanding both typical cognitive function and neurodegenerative disease progression. This dissertation investigates the fidelity and significance of widespread spatiotemporal patterns, specifically Quasi-Periodic Patterns (QPPs) and Complex Principal Component Analysis (cPCA)-derived patterns, in resting-state functional magnetic resonance imaging (rs-fMRI). We evaluate the methodological influences of scan duration, temporal resolution (TR), and frequency filtering on the stability and reliability of these patterns across five independent datasets. Results demonstrate that QPPs and cPCA-derived patterns are highly sensitive to acquisition parameters, with longer scan durations and shorter TRs improving the reliability of extracted patterns. Leveraging these methodological insights, we apply QPP and cPCA analysis to study the longitudinal progression of Alzheimer's disease using rs-fMRI data from clinically characterized cohorts, identifying distinct network disruptions in AD progression. Notably, QPP and cPCA reveal progressive alterations in intrinsic connectivity networks, with disruptions beginning in the default mode network and extending to the frontoparietal, dorsal attention, and subcortical networks. These disruptions are accompanied by increasing phase desynchronization between major brain networks, suggesting a loss of functional coordination over time. Differences in quasi-periodic activity between cognitively stable and transitioning individuals further suggest QPP frequency and global signal alterations as potential early biomarkers for disease progression. By establishing methodological best practices and demonstrating the clinical utility of spatiotemporal dynamics, this work provides a framework for advancing functional neuroimaging biomarkers of Alzheimer's disease.