Purpose:To compare the temporal behaviors of perfusion and blood oxygenation level-dependent (BOLD) functional magnetic resonance imaging (fMRI) in the detection of timing differences between distinct brain areas, and determine potential latency differences between stimulus onset and measurable fMRI signal in sensory cortices.
Materials and Methods:Inversion recovery (IR) spin-echo echo-planar imaging (EPI) and T 2 *-weighted gradient-echo EPI sequences were used for perfusion-and BOLD-weighted experiments, respectively. Simultaneous auditory and visual stimulations were employed in an event-related (ER) paradigm. Signal time courses were averaged across 40 repeated trials to evaluate the onset of activation and to determine potential differences of activation latency between auditory and visual cortices and between these scanning methods.Results: Temporal differences between visual and auditory areas ranged from 90 -200 msec (root-mean-square (RMS) ϭ 134 msec) and from -80 to 930 msec (RMS ϭ 604 msec) in perfusion and BOLD measurements, respectively. The temporal variability detected with BOLD sequences was larger between subjects and was significantly greater than that in the perfusion response (P Ͻ 0.04). The measured time to half maximum (TTHM) values for perfusion imaging (visual, 3260 Ϯ 710 msec; auditory, 3130 Ϯ 700 msec) were earlier than those in BOLD responses (visual, 3770 Ϯ 430 msec; auditory, 3360 Ϯ 460 msec).
Conclusion:The greater temporal variability between brain areas detected with BOLD could result from differences in the venous contributions to the signal. The results suggest that perfusion methods may provide more accurate timing information of neuronal activities than BOLD-based imaging. ADVANCES IN NEUROIMAGING TECHNIQUES, particularly noninvasive magnetic resonance imaging (MRI) methods, have greatly enhanced our understanding of the functional organization of the human brain. Current functional MRI (fMRI) techniques measure regional cerebral hemodynamics (i.e., blood flow or blood oxygenation) to infer information about neuronal activity. Yet, indexing local hemodynamic fluctuations is a poor proxy for neuronal activity and may lead to erroneous conclusions concerning the temporal or spatial nature of neuronal processing. In contrast, human electrophysiologic methods provide precise temporal information about the firing patterns of populations of neurons, though these techniques are limited in their spatial resolution. Auditory and visual evoked potentials typically occur between 10 and 50 msec and 50 and 100 msec after stimulation, respectively (1-3). Here we attempt to replicate these findings using fMRI techniques. Given the electrophysiology results as a gold standard, any differences we find must be associated with either 1) true differences in the physiology or vasculature between these regions or 2) inaccuracies in the imaging methodologies. To assess this second potential, we employed two separate MRI-based functional procedures: blood oxygenation level-dependent (BOLD) and perfu...