Hemodynamic-based inference of cerebral oxygen metabolism

Task-associated neuronal activity induces a complex cascade of biological events, which must balance an increased the supply of oxygen and metabolites carried by the blood to the energetic demands of increased electrophysiology and cellular activity. Although it has been heavily investigated, this cascade of events still remains poorly understood and highly debated. Measurements of cerebral metabolism (such as CMRO2; the cerebral metabolic rate of oxygen) are potentially important towards investigating the interplay between the brain's neural and hemodynamic functions, with implications for the understanding of normal and pathologic neural-metabolic-vascular coupling of the brain. While the ability to measure the net hemodynamic and blood oxygen saturation changes of intravascular blood with methods like functional T2 *-weighted (BOLD- Blood Oxygen Level Dependent) MRI has been invaluable in mapping the functional organization of the brain, due to the dependence of such hemodynamic-based signals on both vascular and metabolic function, these measurements have an ambiguous and potentially non-linear relationship with underlying electrical and metabolic responses; reflecting the balance between neural-metabolic and neural-vascular coupling. Distinguishing between the effects of the vascular and metabolic responses will enable greater quantitative interpretation of hemodynamic signals for the neurosciences, making longitudinal and cross-subject studies more fruitful. The purpose of this work was to develop advanced methods for improving the estimate of CMRO2 from hemodynamic-based imaging methods such as functional MRI and non-invasive optical imaging.; In this work, we introduced and developed model-based data fusion methods for inductively inferring CMRO2 from multimodal optical and fMRI measurements. As a first step to this project, this work included the empirical cross-validation of optical and MR measurement theories (Chapter 5), which led to the development of linear state-space models for multimodal data fusion (Chapter 6) and the development (Chapter 7) and application (Chapters 8 and 9) of a non-linear extension of this fusion method with a multi-compartment vascular state-model. Human cerebral physiology (Chapter 2), functional neuroimaging (fMRI and optical) methods and data analysis techniques developed under this work (Chapter 3), and a review of pervious methods and initial approaches for examining CMRO2 (Chapter 4) will also be discussed as they relate to this project.