| |
Flat Maps
17
rect measure of neural activity in the gray matter [Teo, et al., 1997 #8] since changes in blood
oxygenation levels are measured and not the neural activities directly (Figure 6).
An important question to ask is whether the BOLD signal comes from mainly large blood
vessels. For the retinotopically-organized cortex, it has been found that the signal changes
smoothly. It does not jump discretely between large veins. Therefore, in the retinotopically
organized cortex the signal is caused by the blood flow in large as well as small veins
[Wandell, 1999 #16]. The spatial resolution of a scanner with a 1.5T magnet makes it possible
to distinguish between activations that are 1.5 mm apart.
BOLD fMRI measures the hemodynamic responses. The neural activities, which cause
the hemodynamic responses in the brain, are therefore filtered by the hemodynamic response
function. Although questioned by some researchers, a linear model is often used. In this case,
it is assumed that the fMRI signal is a convolution of a hemodynamic response function with
neural activity [Khan, 1999 #33]. The functional data used for this project come from pre-
processed time series. Sinusoids are fitted to these time series and amplitude, phase, and co-
herence (with the expected response) can be calculated. The sinusoid’s amplitude and phase
can be determined by computing the discrete Fourier transform (DFT) at the stimulus fre-
quency (for example T=36 s, 6 cycles per scan [Khan, 1999 #33]). The detailed procedures are
beyond the scope of this report.
The temporal response of the BOLD fMRI signal in calcarine cortex was evaluated by
Boynton et al [Boynton, et al., 1996 #34]. They modeled the BOLD fMRI signal. One part of
their model was a linear temporal filter for the sluggish vasculature. The delay of the filter was
determined to be about 2.5 s; about 3 s later, a peak is reached. Therefore, for BOLD fMRI,
the temporal resolution is mainly reduced because it takes a few seconds for the oxygenated
blood to accumulate. However, the average temporal resolution (1…2 s) is still better than
that of PET (60 s) [Seminowicz, 2001 #30].
In the visual cortex, the spatiotemporal resolution is therefore sufficient to investigate the reti-
notopic organization of this part of the cortex [Wandell, 1999 #16]. For further information on
fMRI, please refer to the literature, e.g., [Baert, et al., 2000 #35].
|  |
|
| |
|
|
