Date of Award
Doctor of Philosophy (PhD)
The brain is hierarchically organized across a range of scales. While studies based on electrophysiology and anatomy have been fruitful on the micron to millimeter scale, findings based on functional connectivity MRI (fcMRI) suggest that a higher level of brain organization has been largely overlooked. These findings show that the brain is organized into networks, and each network extends across multiple brain areas. This large-scale, across-area brain organization is functionally relevant and stable across subjects, primate species, and levels of consciousness.
This dissertation addresses the neural origin of MRI functional connectivity. fcMRI relies on temporal correlation in at-rest blood oxygen level dependent (BOLD) fluctuations. Thus, understanding the neural origin of at-rest BOLD correlation is of critical significance. By shedding light on the origin of the large-scale brain organization captured by fcMRI, it will guide the design and interpretation of fcMRI studies. Prior investigations of the neural basis of BOLD have not addressed the at-rest BOLD correlation, and they have been focusing on task-related BOLD. At-rest BOLD correlation captured by fcMRI likely reflects a distinct physiological process that is different from that of task-related BOLD, since these two kinds of BOLD dynamics are different in their temporal scale, spatial spread, energy consumption, and their dependence on consciousness.
To address this issue, we develop a system to simultaneously record oxygen and electrophysiology in at-rest, awake monkeys. We demonstrate that our oxygen measurement, oxygen polarography, captures the same physiological phenomenon as BOLD by showing that task-related polarographic oxygen responses and at-rest polarographic oxygen correlation are similar to those of BOLD. These results validate the use of oxygen polarography as a surrogate for BOLD to address the neural origin of MRI functional connectivity.
Next, we show that at-rest oxygen correlation reflects at-rest correlation in electrophysiological signals, especially spiking activity of neurons. Using causality analysis, we show that oxygen is driven by slow changes in raw local field potential levels (slow LFP), and slow LFP itself is driven by spiking activity. These results provide critical support to the idea that oxygen correlation reflects neural activity, and pose significant challenges to the traditional view of neurohemodynamic coupling.
In addition, we find that at-rest correlation does not originate from criticality, which has been the dominant hypothesis in the field. Instead, we show that at-rest correlation likely reflects a specific and potentially localized oscillatory process. We suggest that this oscillatory process could be a result of the delayed negative feedback loop between slow LFP and spiking activity.
Thus, we conclude that at-rest BOLD correlation captured by fcMRI is driven by at-rest slow LFP correlation, which is itself driven by spiking activity correlation. The at-rest spiking activity correlation, itself, is likely driven by an oscillatory process. Future studies combining recording with interventional approaches, like pharmacological manipulation and microstimulation, will help to elucidate the circuitry underlying the oscillatory process and its potential functional role.
Chair and Committee
Lawrence H Snyder
Erik Herzog, Camillo Padoa-Schioppa, Steve Petersen, John Pruett,
Li, Jingfeng, "Neural Basis of Functional Connectivity MRI" (2015). Arts & Sciences Electronic Theses and Dissertations. 550.