5 EEG Source Imaging and Multimodal Neuroimaging
117
of interest. Continuing with Buxton’s model and assuming a neural activity of N(t),
the cerebral blood flow (CBF) and CMRO 2 are respectively modeled as
f (t) 1 + ( f 1 − 1)h
t − δt f
∗ N (t)
(5.60)
and
m(t) 1 + (m 1 − 1)g(t − δt m ) ∗ N (t)
(5.61)
where f 1 and m 1 represent the normalized CBF and CMRO 2 responses, δt f and δt m
are the relative delay of each response from the stimulus, * indicates convolution, and
g(t) and h(t) are impulse response functions. The BOLD signal change can finally
be modeled as:
S
S 0
A ·
1 − f
α−β m
β
(5.62)
where S is the change in MR signal, S 0 is the resting signal, A is the maximum
possible change in BOLD signal (contingent on MRI acquisition factors) and β is a
constant related largely to MRI field strength (~1.5 for 1.5T or 3T machines).
At this point, we have followed a model beginning with a neuronal population and
extending outward to influence both BOLD and EEG signals. While the approaches
presented here have been important to the field, it is important to remember that neurogenerative modeling techniques are highly customizable and different models can
be combined and altered in a number of ways to suit the case at hand. This flexibility
and the explanatory power of the models make them valuable approaches within the
field—particularly when attempting to verify or validate an existing hypothesis.
5.3.3 EEG-FNIRS
fNIRS can be seen as a relative of fMRI. While it does not utilize the strong magnetic
fields of MRI, fNIRS provides researchers with a similar measurement of cortical
hemodynamics. Signals in this case are created by near-infrared light from either laser
diode or LED sources distributed throughout the scalp, which emit light between the
wavelengths of 650 and 950 nm. Light within this range is able to transmit through
the tissues of the head (cortex, skin, skull, scalp, CSF, etc.), though scattering from
these intermediate tissues causes the near-infrared signal to curve through the head in
a banana-shaped pattern. This allows the light achieve a probing depth of ~3 cm and
return back towards the scalp surface, where separate optodes are placed to detect
any changes in the signal intensity (see Fig. 5.11).
As with the magnetic field of MRI, the main focus for detection for fNIRS is
hemoglobin, which serves as the main chromophore in blood and presents with
different absorption spectra depending on whether or not it has bound oxygen. To
117
of interest. Continuing with Buxton’s model and assuming a neural activity of N(t),
the cerebral blood flow (CBF) and CMRO 2 are respectively modeled as
f (t) 1 + ( f 1 − 1)h
t − δt f
∗ N (t)
(5.60)
and
m(t) 1 + (m 1 − 1)g(t − δt m ) ∗ N (t)
(5.61)
where f 1 and m 1 represent the normalized CBF and CMRO 2 responses, δt f and δt m
are the relative delay of each response from the stimulus, * indicates convolution, and
g(t) and h(t) are impulse response functions. The BOLD signal change can finally
be modeled as:
S
S 0
A ·
1 − f
α−β m
β
(5.62)
where S is the change in MR signal, S 0 is the resting signal, A is the maximum
possible change in BOLD signal (contingent on MRI acquisition factors) and β is a
constant related largely to MRI field strength (~1.5 for 1.5T or 3T machines).
At this point, we have followed a model beginning with a neuronal population and
extending outward to influence both BOLD and EEG signals. While the approaches
presented here have been important to the field, it is important to remember that neurogenerative modeling techniques are highly customizable and different models can
be combined and altered in a number of ways to suit the case at hand. This flexibility
and the explanatory power of the models make them valuable approaches within the
field—particularly when attempting to verify or validate an existing hypothesis.
5.3.3 EEG-FNIRS
fNIRS can be seen as a relative of fMRI. While it does not utilize the strong magnetic
fields of MRI, fNIRS provides researchers with a similar measurement of cortical
hemodynamics. Signals in this case are created by near-infrared light from either laser
diode or LED sources distributed throughout the scalp, which emit light between the
wavelengths of 650 and 950 nm. Light within this range is able to transmit through
the tissues of the head (cortex, skin, skull, scalp, CSF, etc.), though scattering from
these intermediate tissues causes the near-infrared signal to curve through the head in
a banana-shaped pattern. This allows the light achieve a probing depth of ~3 cm and
return back towards the scalp surface, where separate optodes are placed to detect
any changes in the signal intensity (see Fig. 5.11).
As with the magnetic field of MRI, the main focus for detection for fNIRS is
hemoglobin, which serves as the main chromophore in blood and presents with
different absorption spectra depending on whether or not it has bound oxygen. To