| dc.description.abstract | In the absence of tasks or external stimuli, the human brain produces recordable spontaneous low-frequency fluctuating hemodynamic signals, which show temporally-correlated activity among brain regions forming a so-called resting-state network (RSN). Multiple RSNs have been identified using functional neuroimaging modalities across species and behavioral states. They are nonstationary over a scan time. These networks collectively cover almost the entire brain and play significant roles in human behaviors. Moreover, abnormalities in RSNs have been identified in major neurological and psychiatric disorders.
RSNs have been studied primarily using the blood oxygenation level-dependent (BOLD) signal generated from functional magnetic resonance imaging (fMRI) machines. Despite numerous studies on these networks, several questions remain partially answered. For example, what is the spatiotemporal structure of networked brain activity obtained from other imaging contrasts such as oxygenated (HbO) and deoxygenated (HbR) hemoglobin signals? Are there observable spatiotemporal differences between these two contrasts? Does the controversial global signal contain neuronally relevant information? This dissertation attempts to answer some of these questions using diffuse optical tomography (DOT). DOT is a non-invasive functional neuroimaging technique based on functional near-infrared spectroscopy (fNIRS) hardware that is more accessible than fMRI, making it available to patients requiring, for example, portability and resistance to head motion.
Several task-based DOT studies have shown a promise in studying the brain’s functional connectivity (FC). However, DOT resting-state studies remain sparse. These studies have imaged limited cortical areas and have assumed that the brain’s FC remains stationary over a scan despite increasing evidence that resting-state FC is a brain-wide nonstationary phenomenon. These limitations were specifically addressed in this dissertation. A significant part of this dissertation involves developing a novel brain-wide DOT (BW-DOT) framework that integrates a cap-based whole-head optode placement system with multiple computational approaches, including finite-element modeling, inverse source reconstruction, data-driven pattern recognition, and statistical correlation tomography, to study brain-wide networked activity in dual contrasts of HbO and HbR. The spatiotemporal resolution of this framework was quantitatively evaluated using simulation and task-based fNIRS experimental data. RSNs were then reconstructed using resting-state data from healthy human participants. The brain’s functional connectivity (FC) was then studied as regions coactivating and deactivating at fine timescales down to a single frame, using the so-called coactivation pattern (CAP) analysis for both HbO and HbR contrasts.
The evaluation results suggest that BW-DOT has the spatiotemporal resolution to study brain-wide networked activity in human adults. Applying BW-DOT to resting-state data revealed multiple RSNs, covering almost the entire neocortical surface. The spatial patterns of these networks suggest statistically significant similarities to fMRI RSN (f-RSN) templates. Furthermore, RSNs obtained from HbO and HbR suggests similarity in the number of RSN types reconstructed and their corresponding spatial patterns. At the same time, HbR RSNs show more similarity to f-RSN templates statistically, and HbO RSNs indicate more bilateral patterns over two hemispheres. In addition, the BW-DOT framework allowed consistent reconstructions of RSNs across individuals and recording sessions, indicating its high robustness and reproducibility, respectively.
The results from CAP analysis show that a small number, i.e., six transient brain-wide CAPs describe major spatiotemporal dynamics of DOT signals. These CAPs represent recurring brain states, have bilaterally symmetrical spatial topographies and exist in highly anticorrelated pairs. Moreover, a structured transitional pattern among six brain states is identified in which two CAPs of anterior-posterior (AP) spatial patterns are in preferred positions, potentially mediating transitions among all brain states. The results further reveal two brain states of global positive and negative patterns spread over the entire cortex. These two global brain states are responsible for generating a subset of peaks and troughs in global signals (GS), supporting the recent reports on the neuronal relevance of hemodynamic GS. Collectively, these results suggest that transient brain events (i.e., CAPs), global brain activity, and brain-wide structured transitions co-exist in humans. They are closely related, extending the observations of similar neuronal events recently reported in animals. The CAP analysis further revealed several differences between the spatiotemporal properties of HbO and HbR DOT CAPs, with the latter contrast having a more structured transitional organization. Overall, this dissertation proposes a novel DOT-based framework and provides insights into the spatiotemporal functional organization of networked brain activity in healthy humans under resting conditions. | en_US |
