Beyond BOLD: Toward the application of quantitative functional magnetic resonance imaging to the study of brain function and physiology

Blood oxygenation level dependent functional magnetic resonance imaging (BOLD fMRI) is a powerful tool for mapping functional anatomy in the human brain. However, the use of BOLD for more than functional mapping is controversial because of the complex relationship between what BOLD measures and the neuronal activity that it purports to reflect. The BOLD signal is primarily a reflection of changes in deoxyhemoglobin in an imaging voxel, which in turn depend on hemodynamic and metabolic changes evoked by neuronal signaling. The magnitude of the BOLD response reflects the balance of the hemodynamic and metabolic responses and is thus difficult to interpret in physiological terms or to compare across groups. The calibrated BOLD approach, which combines BOLD imaging with arterial spin labeling (ASL) and biophysical modeling, has been proposed as a method of untangling the physiological components of the BOLD signal and providing quantitative information about the physiological response to neural activity. This approach has been used successfully in model systems; however, its use remains limited in the fMRI community. In this dissertation I discuss my efforts to address several limitations of this approach with the purpose of broadening its applicability to the study of brain function and physiology: (1) low signal-to-noise ratio (SNR) in ASL-derived blood flow estimates; (2) the need to use CO2 inhalation as a calibration experiment; (3) the limited ability of current biophysical models to describe atypical physiology. To address (1), I developed a method called BOLD constrained perfusion (BCP), which combines information from BOLD and ASL imaging to better estimate blood flow and metabolic fluctuations under conditions where temporal averaging cannot be used to improve SNR. To address (2), I compared traditional CO2 calibration with an alternative approach requiring only the measurement of R2&feet; signal decay. Finally, to address (3), I developed a method of estimating the metabolic response to a stimulus using a detailed model of the BOLD response, then used it to begin investigating the metabolic response to CO2 and visual stimulation in hypoxia.