Biomedical Engineering Reference
In-Depth Information
[68], VEPs [55], SSVEPs, or lateralized readiness potentials [69]. However, some
studies in the past have suggested modified somatosensory evoked potentials and
auditory evoked potentials [70, 71]. Further investigation is needed to analyze the
effects of the scanner environment on spontaneous brain rhythms as well as evoked
potentials and event-related potentials generated by different sensory modalities.
12.2.3 Using fMRI to Study EEG Phenomena
Since the first recording of the EEG by Hans Berger in the 1920s, efforts have been
made to understand the significance and the genesis of various spontaneous
rhythms as well as induced patterns in the EEG. These include the well-known
alpha (8 to 12 Hz) and beta (16 to 25 Hz) rhythms, sleep spindles (12 to 16 Hz),
interictal epileptic spikes, as well as evoked and event-related potentials. In most of
these cases, inverse modeling approaches have been used to estimate the location of
the generators of EEG current dipoles. However, due to volume conduction and the
mixing effects of multiple dipoles, the localization of sources of EEG activity cannot
be determined uniquely or with a high spatiotemporal resolution. Furthermore,
these methods require strong a priori assumptions about the head model and
impose restrictions on the location and number of possible dipoles, most of which
are not easily verifiable [72]. fMRI, on the other hand, allows for imaging the func-
tional activation of various anatomical regions of the brain with a high spatial reso-
lution, including deeper structures such as the cerebellum and midbrain.
Increasingly, fMRI is being used to investigate EEG phenomenon, such as brain
rhythms, evoked potentials, or event-related potentials as well as conditions that
can be monitored or classified using features of the EEG, such as epilepsy, sleep
stages, or the like.
12.2.3.1 fMRI Correlates of EEG Rhythms
In recent years various methods have been suggested for studying the relationship
between fMRI and rhythms of the EEG. The most widely studied of these is the pos-
terior alpha rhythm. The basis of these methods is to correlate the time course of the
power of this frequency band (as observed in each EEG channel) with the BOLD
response of each voxel. Power spectra are calculated using standard spectral estima-
tion techniques such as the short time Fourier transform or wavelet analysis, fol-
lowed by convolution with a universal hemodynamic response function (that
characterizes the coupling of neuronal activation to the BOLD response). This time
series is then correlated with the observed BOLD response of each voxel of the brain
[50, 73-75]. Figure 12.13(a) depicts the results of such a study, showing correlation
between spontaneous fluctuations of the posterior alpha rhythm and the BOLD sig-
nal between various regions of interest in the brain. Figure 12.13(b) shows the
functional images obtained by this method.
Although this is simple and intuitive, each stage of this technique has been mod-
ified by other groups to overcome some of the drawbacks, not obvious at first
glance. For instance, the EEG recorded by each channel is neither an exclusive result
of the activity directly underneath the corresponding scalp electrode nor reflective
of independent neuronal phenomenon. Each channel of the EEG is rather a
Search WWH ::
Custom Search