Abstract
Perfusion based event-related functional MRI (ER-fMRI) has been recently developed to study human brain activation in response to brief stimulation, since the information of cerebral blood flow measures, as a direct estimation of hemodynamic responses, may provide better localization of brain activation, and improve our understanding of underlying physiological mechanism. In one brain region, temporal resolving power of fMRI is limited by the smoothed hemodynamic response function dispersed from the neuronal activity. In this work, temporal modulation transfer functions were consequently utilized to quantify the resolving powers of perfusion and blood oxygenation level-dependent (BOLD) fMR signals in time domain. The impulse response function was determined using brief visual stimulations and event-related image acquisition schemes. An important feature of arterial spin l ling techniques is that quantitative perfusion and BOLD signals could be simultaneously acquired. This simultaneous BOLD response may arise from signals that are more proximal to capillary beds, and its temporal resolution may be different from that of the typical BOLD response. Therefore, we assessed and compared the temporal resolving capabilities of perfusion, simultaneous BOLD, and the typical BOLD response obtained from the gradient echo EPI pulse sequence. Full-width-at-half-maximums of perfusion and simultaneous BOLD measurements were significantly smaller than that of BOLD ones (4.3±0.6 sec vs. 5.5±0.9 sec, p<0.02 and 4.5±0.7 sec vs. 5.5±0.9 sec, p<0.01, respectively). The corresponding temporal resolving powers of perfusion and simultaneous BOLD signals were statistically better than that of BOLD signals (0.23±0.03 Hz vs. 0.17±0.02 Hz, p<0.01 and 0.22±0.03 Hz vs. 0.17±0.02 Hz, p<0.01, respectively). Our results showed that the typical BOLD response was significantly smoothed from the perfusion response, thus resulted in a degraded temporal resolving power. However, results from the simultaneous BOLD and perfusion measurements were not significantly different. Biophysical implications of the experimental outcomes were further investigated using a computer simulation based on the Balloon model. By fitting the measured data into the model, an apparently longer transit time was obtained for the typical BOLD signal (1.7 sec), comparing to that for the simultaneous BOLD one (0.8 sec). Therefore, the simultaneous BOLD signal was regarded as less susceptible to the variations from local draining veins. Combining the simulation result with the significantly discrepant resolving powers between the two BOLD signals, we speculated that the blurred effects from large vessels played a predominant role that further reduced the temporal resolution of the BOLD-based fMRI from the perfusion response. Then, we have also compared the accuracy in the estimation of onset times of the perfusion and BOLD based ER responses between two brain regions. Simultaneous auditory and visual stimulation that was with ideal temporal difference < 100 ms at cortices was employed in an ER paradigm design with 40 repeated single trials. In order to provide corresponding signal-to-noise ratio (SNR) of blood flow and oxygenation sensitive measures in the comparison between distinctive brain areas, slice-selective inversion recovery (IR) and BOLD-contrast GE EPI sequences were used for the perfusion- and BOLD- sensitive imaging, respectively. Temporal discrepancies between the two regions ranged from 90 to 200 ms (RMS = 134 ms) in IR, and from -80 to 930 ms (RMS = 604 ms) in BOLD measurements. Onset differences detected in the BOLD response varied significantly among subjects and were statistically greater than that in the IR response (p < 0.04), which may be resulted from differences in the venous contributions in different regions. Mention when the temporal resolving abilities in a single brain region assessed utilizing temporal MTF or that between two brain regions evaluated using the onset differences, it was suggested that perfusion-based ER-fMRI, being capable of localizing better to the brain parenchyma, provided more accuracy timing information of neuronal activities, comparing to the BOLD signals.