How are BOLD Measures and Connectivity Affected by Noise Removal in Resting-state and Task fMRI

Functional magnetic resonance imaging (fMRI), based on blood-oxygenated-level dependent (BOLD) contrast, is a powerful non-invasive tool to assess brain function and connectivity in vivo. However, as BOLD signals are noisy, properly recovering the signal of interest from noise-related uctuations is essential to obtain reliable measures. To this end, several pre-processing pipelines have been developed, mainly adopting nuisance regression and independent component analysis (ICA). While previous works assessed the impact of these cleaning methods on resting-state fMRI (rs-fMRI) time-series and focused on specic aspects of signal cleaning (e.g. motion removal), their wider effect on both resting-state and task fMRI is yet to be fully investigated. In this work, we aimed to assess the inuence of different pre-processing methods on BOLD measures, by considering a rs-fMRI dataset, where no a priori information regarding which brain areas are expected to be active is present, and a task fMRI dataset, for which there is a strong hypothesis regarding active brain areas