Introduction
Cerebral hemodynamics plays a very important role in the delivery of oxygen to brain tissue. The objective of this work is to estimate the dynamic changes in CMRO2 that result from evoked changes in neuronal activity using CBF and BOLD fMRI data and also to determine how closely coupled are the changes in CMRO2 to the changes in neuronal function.
Methods
The model of the regional hemodynamics consisted of multiple parts that represent the various aspects of the hemodynamic response to a stimulus1, 2, 3, 4, 5. The coupling between the CMRO2 and hemodynamic responses was examined by equating it as a first-order process from the neuronal response (akin to an intermediary process). A two-echo gradient-echo FAIR acquisition scheme was used to obtain simultaneous CBF and BOLD fMRI data with temporal resolution of 2 s in a General Electric 3.0 T MRI scanner 6 . Study participants were instructed to perform a visually cued finger tapping task with stimulation periods of 12 s followed by 38 s of rest.
Results and Discussion
The CMRO2 response was represented by a time constant (τ) and two scenarios were considered: 1 the CMRO2 response is as fast as the neuronal response (τ=0), 2 the CMRO2 response is slower than the neuronal response (τ>0). The results showed that a slower CMRO2 response reduced the NMSE (normalized mean squared error) of the CBF and BOLD predictions versus a fast CMRO2 response by 20% (Fig. 1). A fast CMRO2 response produced larger NMSE estimates that included non-physiologic characteristics. These estimates suggest the CMRO2 response is not as fast as the neuronal response, but moderately faster than the blood flow response (Fig. 1 right). These results are limited by the models and assumptions used and our current knowledge of the cerebral hemodynamics but these models can be used to increase our understanding of the hemodynamic mechanisms and exploit the information in the BOLD response under different physiological conditions.