FIGURE 13.1. Modeling fMRI signals. A. In response to a single brief impulse of neural activity, the fMRI BOLD response lags the neural activity by about 5 seconds and is characterized by 3 epochs: a) the initial dip, b) the positive hemodynamic response, and c) the postresponse undershoot. B. Hypothetical neural activity during a delayed-response task, where C is a cue to be remembered, and R is the response occurring after an imposed delay. The evoked fMRI BOLD response involves a mixture of signals emanating from more than one time and more than one trial component. The gradient under the curve schematically represents the mixing or temporal overlap of the various signal components. Whiter regions reflect purer (less colinear) BOLD signal, and darker regions reflect highly colinear signal. For example, the white region at the peak of the first hump is almost exclusively evoked from neural processing during the cue phase of the task. However, just a few seconds later, in the darker portion just to the right, the signal is a mixture of processing at the cue phase and the beginning of the delay period. C. To resolve the individual components of the mixed fMRI signal, ideal hemodynamic response functions (which take into account the lag and spread of the BOLD response) are used to model within-trial components. In this case, a separate covariate is used to model the cue, delay, and response phase of the trial. D. The covariates are entered into the modified GLM of the fMRI time-series data, and a least-squares procedure is used to derive parameter estimates (i.e., beta values) that scale with the degree to which a given covariate accounts for the variance in the observed data. For example, the height of the delay covariate can be used as an index of the amount of delay-period activity.
First, in response to transient increases in neuronal oxygen consumption, the BOLD signal decreases because the ratio of oxy-/deoxyhemoglobin in blood decreases. This transient decrease has been termed the initial dip and is currently under increased scrutiny because it may provide greater spatial localization than subsequent responses (Ugurbil et al., 1999; Yacoub & Hu, 2001). Second, a large increase in signal above baseline is observed beginning approximately 2 seconds and peaking 4 to 6 seconds after the onset of a brief impulse of neural activity (the precise latency and time course of the response can vary depending on the individual, the brain region, or the length of the neural activation; Aguirre, Zarahn, & D'Esposito, 1998; Buckner,
1998). This increase is caused by a local increase in blow flow that actually overcompensates for the amount of oxygen consumed. Thus, the ratio of oxy-/deoxyhemoglobin increases in the vasculature near the site of neural activity. Most fMRI studies primarily focus on this positive phase. Finally, there is a decrease in signal that most often falls below baseline and can require tens of seconds to return to baseline.
Was this article helpful?