From Scans to Reality: Effects of Preprocessing and Daily Behavioral Patterns on fMRI Brain Connectivity

Functional magnetic resonance imaging (fMRI) has significantly advanced our understanding of the brain. By combining data from hundreds of individuals, neuroscientists have been able to reliably identify general trends of the large-scale network organization of the brain in controls and patients. Consequently, there is optimism that fMRI connectivity could be used to identify biomarkers for prognosis and diagnosis and targets for interventions. However, the clinical application of fMRI techniques remains limited, partly because of the heterogeneity in findings from between-group comparisons and the challenge of group-to-individual generalization. Therefore, it is critical to investigate and understand the underlying factors contributing to these challenges, which might include fMRI preprocessing and the impact of environmental and behavioral neuromodulators on a day-to-day basis. In this thesis, I explore these sources of variability to understand their impact on fMRI brain connectivity. This includes the impact of preprocessing methods and the role of environmental and behavioral neuromodulators -- or external factors -- as well as examining the integration of fMRI with digital phenotyping to provide a comprehensive understanding of the brain. In the first study, I demonstrate that spatial smoothing has unpredictable effects when comparing resting-state fMRI data between groups. In the second study, I focus on developing software tools for preprocessing, analyzing, and visualizing digital phenotyping data, aiming to quantify the influence of external factors more effectively. In the third study, I review how the combination of brain MRI and digital phenotyping devices enables the monitoring of external factors in real-world scenarios, offering a more precise method to quantify these variables. Finally, in the fourth study, I show that external factors, including sleep, physical activity, mood, respiration rate, and heart rate variability, are significantly correlated with functional connectivity, a relationship that persists up to fifteen days prior. This thesis demonstrates that both preprocessing choices and external factors uniquely influence fMRI brain connectivity, often in unexpected ways. It supports merging digital phenotyping with MRI data to bridge brain research and real-life experiences, extending brain research from scanners to reality. My findings reveal temporal co-variations between external factors and brain connectivity, crucial for understanding mental health disorders that exhibit week-to-week variability. Thus, integrating brain connectivity analysis with insights into environmental and behavioral neuromodulators propels environmental neuroscience forward and supports the development of precision healthcare, making significant strides towards personalized medicine and the understanding of individual variability.