integration of fmri and simultaneous eeg: towards a
TRANSCRIPT
www.elsevier.com/locate/ynimg
NeuroImage 22 (2004) 83–94
Integration of fMRI and simultaneous EEG: towards a
comprehensive understanding of localization and time-course
of brain activity in target detection
Christoph Mulert,a,* Lorenz Jager,b Robert Schmitt,a Patrick Bussfeld,a Oliver Pogarell,a
Hans-Jurgen Moller,a Georg Juckel,a and Ulrich Hegerla
aDepartment of Psychiatry, LMU, Munich, Germanyb Institute of Clinical Radiology, LMU, Munich, Germany
Received 6 May 2003; revised 29 October 2003; accepted 29 October 2003
fMRI and EEG are complimentary methods for the analysis of brain
activity since each method has its strength where the other one has
limits: The spatial resolution is thus in the range of millimeters with
fMRI and the time resolution is in the range of milliseconds with EEG.
For a comprehensive understanding of brain activity in target
detection, nine healthy subjects (age 24.2 F 2.9) were investigated
with simultaneous EEG (27 electrodes) and fMRI using an auditory
oddball paradigm. As a first step, event-related potentials, measured
inside the scanner, have been compared with the potentials recorded in
a directly preceding session in front of the scanner. Attenuated
amplitudes were found inside the scanner for the earlier N1/P2
component but not for the late P300 component. Second, an
independent analysis of the localizations of the fMRI activations and
the current source density as revealed by low resolution electromag-
netic tomography (LORETA) has been done. Concordant activations
were found in most regions, including the temporoparietal junction
(TPJ), the supplementary motor area (SMA)/anterior cingulate cortex
(ACC), the insula, and the middle frontal gyrus, with a mean Euclidean
distance of 16.0 F 6.6 mm between the BOLD centers of gravity and
the LORETA-maxima. Finally, a time-course analysis based on the
current source density maxima was done. It revealed different time-
course patterns in the left and right hemisphere with earlier activations
in frontal and parietal regions in the right hemisphere. The results
suggest that the combination of EEG and fMRI permits an improved
understanding of the spatiotemporal dynamics of brain activity.
D 2004 Elsevier Inc. All rights reserved.
Keywords: fMRI; EEG; ERP; P300; Auditory-evoked potential; Target
detection; Oddball
Introduction
For years, research in auditory target detection has been focused
on event-related potentials. Using an oddball paradigm, a large
positive component can be recorded after about 300 ms (the P300
1053-8119/$ - see front matter D 2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.neuroimage.2003.10.051
* Corresponding author. Department of Psychiatry, LMU, Nußbaum-
straße 7, 80336 Munchen, Germany. Fax: +49-89-5160-5542.
E-mail address: [email protected] (C. Mulert).
Available online on ScienceDirect (www.sciencedirect.com.)
potential), if a rare, task-relevant target has been presented. This
component can be reliably found with largest amplitudes in parietal
electrodes on the scalp. Major interest to this potential was caused
by the finding that in several neuropsychiatric disorders, like
dementia of the Alzheimer type or schizophrenia, attenuations of
the amplitude and latency have been described (Frodl et al., 2002a;
Hegerl et al., 1995; Polich and Pitzer, 1999; Winterer et al., 2001).
However, the clinical benefit has been limited so far, since the
finding of an attenuated P300 amplitude seems to be rather
unspecific when looking at scalp-data. The topography on the
scalp allows apparently only a rough distinction between a more
frontally pronounced early P3a component and a more parietally
pronounced P3b component. The P3a has been described to be
involved in automatic novelty detection and the P3b is more
concerned with volitional target detection (Soltani and Knight,
2000). Using intracranial measurements, however, it could be
demonstrated that a complex network of brain regions is involved
in the generation of the P300 potential, including temporoparietal
junction, posterior superior parietal regions, cingulate cortex,
dorsolateral prefrontal cortex, ventrolateral prefrontal cortex, and
medial temporal regions (Kiss et al., 1989; McCarthy and Wood,
1987). Interestingly, these regions also show different time-course
patterns (Baudena et al., 1995; Halgren et al., 1995a,b).
With regard to the meaning of a disturbed P300 in brain
diseases, it became clear that the important next step would be to
get similar information with high spatial and time resolution
noninvasively, since intracranial measurements are usually not
appropriate in a clinical setting. With the development of nonin-
vasive hemodynamic-based brain imaging methods like fMRI,
with its high spatial resolution in the range of millimeters, several
studies have been undertaken to figure out brain regions involved
in auditory target detection (Downar et al., 2000; Kiehl et al., 2001;
Linden et al., 1999). Interestingly, beside the medial temporal
region, all other brain regions that have been known from the
intracranial measurements could be detected again using fMRI,
suggesting that there is a high degree of concordance between the
electrical signal and hemodynamic response. Concerning the
disturbed P300 potential in diseases like schizophrenia, it was
evident that a noninvasive investigation of disturbed brain function
C. Mulert et al. / NeuroImage 22 (2004) 83–9484
in target detection became possible now with high spatial resolu-
tion (Kiehl and Liddle, 2001).
However, the intracranial measurements by Halgren et al. had
revealed clear differences in the time-course patterns of the
involved brain regions with latency peak differences in the range
of 100 ms and less. Therefore, it was also clear that hemodynam-
ic-based methods would not have a sufficient high temporal
resolution for a distinct analysis of the timing of brain activity
in target detection. Thus, a combination of electrical activity- and
hemodynamic-based methods is suggestive since these methods
are complementary because of the good temporal resolution of
EEG/ERP-based approaches and the good spatial resolution of
fMRI. For such a combination approach, it would be critical that
the signals generated by each method correspond to the same set
of underlying generators (Horwitz and Poeppel, 2002). The good
concordance of the intracranial measurements and the fMRI
studies did already encourage this approach. Concerning basic
research, important support came from recent physiological stud-
ies in monkeys (Logothetis et al., 2001), investigating simulta-
neously intracortical recordings of electrical neural signals and the
blood-oxygen-level-dependent (BOLD) fMRI-responses in the
visual cortex. Clear correlations were found between the local
field potentials, which are mostly reflecting a weighted average of
synchronized dendro-somatic components of the input signals of a
neural population, and the BOLD responses. No such clear
relationship could be found between multiunit spiking activity
and the BOLD signal. Since the local field potentials are the
physiological basis of the scalp EEG, this result is in favor to the
hypothesis that the signals generated by EEG and fMRI might
correspond to the same set of underlying generators.
In principle, there are three possibilities: Cerebral activity might
be detectable only with EEG, only with fMRI, or with both
methods. The first might be true if, for example, short-lasting
neuronal activity, leading to a short ERP component, does not
evoke a measurable (and significant) hemodynamic response. The
second could occur if long-lasting neuronal activity would result in
a large hemodynamic response but maybe only in a slow shift of the
electrical baseline that is not detected in the classical peak picking
approach. Finally, brain activity should be detectable with both
methods in the case of a synchronous activity of a large number of
neurons for some hundred milliseconds. Then, probably both a large
ERP component and a significant BOLD signal would be expected.
This could be true for the P300 component: An important study by
Horovitz et al. (2002) could already show a close relationship
between the scalp-P300 and the BOLD signal in an oddball
paradigm. Manipulation of the task (probability of the occurrence
of a target) was found to influence both the amplitude of the P300-
component and the BOLD-signal changes in most of the involved
brain regions in the same direction. So, an experimental condition
that did result in an increased P300 amplitude at Pz did also result in
an increased BOLD-signal change (e.g., in the supramarginal gyrus)
and vice versa (Horovitz et al., 2002). EEG and fMRI measurement
was done in separate sessions in this experiment.
Due to the technical challenge of a simultaneous measurement
of EEG and fMRI, a combination of EEG and fMRI so far has
mostly been done measuring with each method in separate sessions.
Using such an approach, many additional variables are convoluted,
such as attention, vigilance, familiarity with a task, which might not
be the same in two separate data acquisition sessions, or experi-
mental environment (lying inside a noisy and narrow scanner with
dim lightning, or sitting comfortable in a quiet EEG lab with natural
light). That in fact the arousal level of a subject is crucial for the
activation of brain regions in cognitive tasks has recently be
demonstrated by Matsuda et al. using a simultaneous EEG/fMRI
study. Here, the authors could find significant activations in a
smooth-pursuit eye movement task in brain areas like the parietal
eye field and the supplementary eye field only during the high
arousal level (Matsuda et al., 2002). Beside the possible implica-
tions of this finding in studies dealing with the comparison between
healthy volunteers and patients with potentially disturbed vigilance/
arousal dynamics, it becomes clear that vigilance and arousal is a
variable that could influence neural activity and therefore should be
controlled if the interpretation of the results shall not be extremely
limited. The only way to make results comparable in a way that
differences between the results obtained by EEG or fMRI are based
on the physiology and on fundamental properties of the methods
and not on confounding variables would be to measure simulta-
neously with both methods. Thus, to get usable and interpretable
results, it is worthwhile to make more effort concerning methodol-
ogy. Simultaneous measurement of EEG and fMRI requires a
special EEG hardware that can be used in the MRI scanner without
making artifacts and with no safety risk. These efforts have been
primarily undertaken in the analysis of epileptic activity (Jager et
al., 2002). Obviously, epileptic activity is in fact a single event that
cannot simply be repeated on demand and therefore a simultaneous
measurement with both EEG and fMRI is especially important.
However, the abovementioned influence of vigilance or arousal on
activations in cognitive fMRI studies make a simultaneous data
acquisition also necessary in target detection. We have therefore
established an experimental set-up that allows a simultaneous
measurement of EEG and fMRI data.
With this study, it was intended to answer three questions: (1) Is
the EEG-signal (P300 potential), as measured inside the MR
scanner, comparable to the signal outside the scanner, measured
in a directly preceding session? (2) Are the same localizations for
activated brain regions found with fMRI and EEG signals? (3) Do
the involved brain regions differ with respect to the time-course of
the neuroelectric activity?
Methods
Subjects
Ten healthy volunteers with no history of neurological or
psychiatric disturbance or reduced hearing were recruited from
an academic environment. One data set was later excluded due to
technical problems. Finally, the data of nine subjects (six men,
three women: range, 20–30 years old; mean age, 24.2 F 2.9) were
analyzed. The study was approved by the local ethics committee of
the Ludwig-Maximilians-University of Munich and written in-
formed consent was obtained from each subject.
Paradigm
Auditory stimuli were generated on a Personal Computer using
the BrainStim software package (Brain Products, Munich, Ger-
many) and conducted through a pair of plastic tubes into a set of
headphones placed over the subjects ears. The auditory stimuli
were a 800-Hz sinus-tone (nontarget, 80%) and a 1300-Hz sine-
wave tone (target, 20%) at a sound pressure level of 95 dB. The
tones were presented with an interstimulus of 3 or 6 s (mean: 4 s).
Fig. 1. Comparison between the auditory-evoked potentials recorded inside and outside the scanner. (a) Clearly attenuated N1/P2 potential inside the scanner (at
Cz). (b) Only slightly reduced P300 potential inside the scanner (at Pz). Outside the scanner, a pronounced P3a component is evoked.
Fig. 2. Comparison between target and nontarget. (a) Inside the scanner: neither the target nor the nontarget stimulus evokes a pronounced P3a component. (b)
Outside the scanner: both the target and nontarget stimulus evoke a P3a component.
C. Mulert et al. / NeuroImage 22 (2004) 83–94 85
C. Mulert et al. / NeuroImage 22 (2004) 83–9486
Targets appeared in pseudo-random order. The volunteers had to
press a button with the right index finger.
Acoustic environment, sound system, and headphones
For the measurement of the sound pressure level inside and
outside the scanner, we have used the following equipment: A 1/2-
in. microphone (B&K 4313, Bruel and KJAER, Denmark) attached
with a preamplifier (B&K 2639) with extension cable (B&K A027)
connected to a measuring amplifier (B&K) served as sound level
recording unit. To enable offline recording, the AC output of the
measuring amplifier was connected to the line input of a digital
audio tape (DAT) recorder (SONY 300 ES, Japan). The measuring
amplifier and the DAT recorder were placed outside the shielded
MR room and the microphone cable was routed via an opening
shielded by copper tubes to suppress radio frequency (RF) induc-
tion. Before the recordings, a calibration procedure was performed
using an acoustic calibrator (B&K 4230). The signals were
recorded at the position of the human head in the isocenter of
the MR system and the second one, at 1.2 m in front of the opening
of the MR. Due to the impulse characteristic of the MR noise, a
large difference between averaging measurements applying ‘‘fast’’
time constant and peak detection occurred. The echo planar
imaging (EPI) sequence used in our study produced a sound
measured in sound pressure level (SPL) of 90.5 dB (dB = decibel)
(fast) and of 106.5 dB (peak) inside the MR system. The basic
Fig. 3. Global field power of the target-specific activity (target minus nontarget)
highly corresponding. The target-specific brain activity is even more pronounced
noise (due to the vacuum pump) inside the MR system was 76 dB.
At 1.2 m in front of the opening of the MR, a basic noise of 62 dB
was measured. Binaural sound transmission was performed using
an air tubing sound delivery system (Resonance Technology, Inc.
Van Mays, USA). The personal computer was placed outside the
shielded MR room. The sound delivery system was evaluated
using standardized equipment (microphone 4144, preamplifier
2619, calibrator 4230, and artificial ear with 2-cm coupler 4152;
Bruel and KJAER, Denmark). Because we combined an invert
earphone with circum-aural ear muffs, a high attenuation against
the noise of the MR (20 dB at low and 50 dB in the high
frequencies) occurred. Due to the tube-based sound transportation
and the sound driver characteristics, a rippled band pass frequency
response occurred. Therefore, the desired level of 95 dB SPL was
achieved by a digital attenuation to the correct intensity.
fMRI setup
Imaging was performed in a 1.5 T Siemens Magnetom Vision
scanner with echo planar capability. For functional BOLD imag-
ing, 10 slices with a T2-weighted EPI sequence (TR: 3s; TE: 64
ms; FOV: 210/280 mm; matrix: 128 � 64; interleaved slice
acquisition; slice thickness: 8 mm; interslice gap: 0.8 mm; result-
ing pixel size: 3.16 � 2.11 mm) were acquired in the same position
as the anatomical images. Each functional time series consisted of
120 volumes each. Five functional time series were acquired. The
inside the scanner and outside the scanner. The shapes of both curves are
inside the scanner.
Fig. 4. Average (n = 9) three-dimensional maps of BOLD-signal increase
for targets versus nontargets ( P < 0.05, corrected for multiple testing) on
the inflated hemispheres of a single subject. Lateral views.
Fig. 5. Average (n = 9) three-dimensional maps of BOLD-signal increase
for targets versus nontargets ( P < 0.05, corrected for multiple testing) on
the inflated hemispheres of a single subject. Medial views.
C. Mulert et al. / NeuroImage 22 (2004) 83–94 87
first two functional image frames of each time series were
discarded to allow for signal equilibration, giving a total of 590
frames used in analysis. For each subject, a three-dimensional
MPRAGE data set (T1-weighted) was acquired. The subject’s head
was immobilized during the whole experiment to minimize invol-
untary head movements.
Post-processing
The complete post-processing of the data including motion
correction, statistics, and transfer of the data into Talairach space
was done with the Brainvoyager software package, version 4.9
(Rainer Goebel, Maastricht, Netherlands). The first two scans of
each run have been excluded from further analysis because of T1
saturation effects. Motion correction was done using Trilinear
Interpolation on a reduced (12.5%) data set. Data smoothing was
performed as temporal data smoothing and linear trend removal.
No spatial data smoothing has been performed.
Functional data were transformed into Talairach space, aligned
with the three-dimensional anatomical volumes from the same
session, and interpolated to a resolution of 1 mm3. For further
statistical analysis, multiple regression analysis was performed on
the three-dimensional functional volume time courses (5 for every
subject with 118 time points each) using the General Linear Model
(GLM). The design matrix contained the idealized response func-
tions that corresponded to the signal time-courses (target, nontar-
get). For statistical analysis, any combination of predictors can be
used. Statistical maps were computed for analysis of individuals
(contrast: target vs. nontarget) and thresholded at P < 0.05 (cor-
rected for multiple testing). Additionally, to obtain information
about activation that are consistent for the entire group, a statistical
group analysis was performed, thresholded at P < 0.05 or P < 0.001
(corrected for multiple testing), respectively (see figures and tables).
EEG
We used an EEG amplifier that cannot be saturated by any MR
activity (EMR; Schwarzer, Munich, Germany) and which has been
specifically developed for operation in the MR system. The EEG
recording was performed with 27 electrodes (sintered silver and
silver-chloride) referred to Cz and placed on the scalp according to
the international 10–20 system with the additional electrodes FC1,
FC2, FC5, FC6, T1, T2, CP5, CP6. Electrode skin impedance was
usually less than 5 kV. Data were collected with a sampling rate of
500 Hz and an analogous band-pass filter (0.16–30 Hz). Two
hundred milliseconds of prestimulus baseline and 575 ms post-
stimulus periods were evaluated for every sweep. For artifact
suppression, all sweeps were automatically excluded from averag-
ing when the voltage exceeded 100F AV in any of the 27 channels
at any point during the averaging period. For each subject, the
remaining sweeps were averaged after baseline correction. Only
wave-shapes, based on at least 40 averages, were accepted. The
Fig. 6. fMRI activations ( P < 0.05, corrected for multiple testing) and
EEG–LORETA current source density activations at corresponding slices
(Z = 8, 15, 22). Good correspondence in the temporoparietal region, but not
in subcortical regions (not covered by LORETA).
Fig. 7. fMRI activations ( P < 0.05, corrected for multiple testing) and
EEG–LORETA current source density activations at corresponding slices
(Z = 29, 36, 44). Good correspondence in the midline and the right middle
frontal gyrus.
C. Mulert et al. / NeuroIma88
EEG recording was acquired and displayed using the BrainLab
software (OSG, Rumst, Belgium). All data processing (artifact
rejection and averaging) has been done using the BrainVision
Analyzer software (Brain Products).
Peak detection, statistics
Peak detection has been done semiautomatically using the
Analyzer software. The N1 peak has been defined as the most
negative value between 80 and 150 ms, the P3 peak as the most
positive value between 250 and 500 ms poststimulus. Differences
between N1 and P3 peaks inside and outside the scanner have been
compared with paired t tests using the SPSS-software (SPSS 11.5).
Parametrization
According to the global field power time course (difference
wave corresponding to the infrequent tone minus frequent tone),
the timeframe 250–540 ms poststimulus has been chosen for
further analysis (Fig. 3).
LORETA
Low resolution electromagnetic tomography (LORETA)
assumes that the smoothest of all activity distributions is most
plausible (‘‘smoothness assumption’’) and therefore, a particular
current density distribution is found (Pascual-Marqui et al., 1994).
This fundamental assumption of LORETA directly relies on the
neurophysiological observation of coherent firing of neighboring
cortical neurons during stimulus processing (Gray et al., 1989;
Llinas, 1988; Silva et al., 1991) and therefore can be seen as a
physiologically based constraint. However, this coherent firing has
been described on the level of cortical columns, which have a
much smaller diameter than the voxels used in the LORETA
software; the empirical basis for coherent firing in the millimeter
range is not strong enough to fully accept this constraint as a
physiological one, even if it might help to produce useful results.
The characteristic feature of the resulting solution is its relatively
low spatial resolution, which is a direct consequence of the
smoothness constraint. Specifically, the solution produces a
‘‘blurred-localized’’ image of a point source, conserving the
ge 22 (2004) 83–94
Fig. 8. LORETA current source density maxima (green squares) and fMRI statistical maps ( P < 0.05, corrected for multiple testing) superimposed on an
anatomical MRI image. Good correspondence in insula, temporoparietal junction (TPJ), and bilateral middle frontal gyrus. The middle and inferior temporal
gyri have not been covered by the fMRI slices.
C. Mulert et al. / NeuroImage 22 (2004) 83–94 89
location of maximal activity, but with a certain degree of disper-
sion. It should be emphasized that this solution will typically
produce a ‘‘blurred-localized’’ image of arbitrary distributions
due to the principle of superposition. However, some distributions
of point sources may superpose in such a way that they actually
cancel out on the scalp and therefore cannot be correctly localized
by any method. The version of LORETA used in the present study
used the digitized Talairach atlas (Talairach and Tournoux, 1988)
available as digitized MRI from the Brain Imaging Center, Mon-
treal Neurologic Institute, estimating the current source density
(mA/mm2) distribution for either single time-points or epochs of
brain electric activity on a dense grid of 2394 voxels at 7-mm
spatial resolution (Pascual-Marqui et al., 1999). The solution space
Table 1
Corresponding activations (target minus nontarget) in fMRI and EEG
Region fMRI Number of E
X, Y, Z (Talairach)activated
voxelsX
SMA/gyrus cinguli 0, �10, 43 7268 (**) �TPJ left �49, �27, 32 2301 (**) �TPJ right 57, �24, 19 3214 (**) 6
Insula left �38, �1, 7 7973 (**) �Insula right 42, 4, 6 6777 (**) 5
STG left �42, �24, 12 2390 (**) �Lobulus parietalis
inferior left
�39, �43, 46 1286 (**) �
Middle frontal
gyrus right
37, 36, 23 104 (**) 5
Precuneus �5, �70, 46 165 (*) �Middle frontal
gyrus left
�30, 39, 25 48 (*) �
SMA = supplementary motor cortex; TPJ = temporoparietal junction; STG = supe
(corrected for multiple testing); (*) number of voxels significant activated on P va
(the three-dimensional space where the inverse EEG problem is
solved) was restricted to the gray matter and hippocampus in the
Talairach atlas (anatomically based constraint). Localization
concerning spherical and realistic head geometry was done using
EEG electrode coordinates reported by Towle et al. (1993). A voxel
was labeled as gray matter if it met the following three conditions:
its probability of being gray matter was higher than that of being
white matter, its probability of being gray matter was higher than
that of being cerebrospinal fluid, and its probability of being gray
matter was higher than 33% (Pascual-Marqui et al., 1999). LOR-
ETA has been widely used in the last years to localize electrical
generators of scalp EEG data (Gallinat et al., 2002; Mulert et al.,
2001, 2002, 2003; Park et al., 2002; Pizzagalli et al., 2001).
EG (LORETA) Current source density Euclidean
, Y, Z (Talairach)[AV/mm2 (�10�3)] distance (mm)
3, �11, 64 5.21 21
59, �25, 22 3.90 14
0, �18, 15 3.17 7
38, 17, 1 2.69 17
3, 10, 8 3.82 12
59, �32, �6 3.57 26
45, �46, 50 4.19 8
3, 17, 22 3.47 24
3, �53, 57 4.96 20
31, 38, 36 4.08 11
rior temporal gyrus; (**) number of voxels significant on a P value < 0.001
lue < 0.05 (corrected for multiple testing).
Table 2
No correspondence between activations (target minus nontarget) in fMRI
and EEG
Region fMRI Number EEG (LORETA) Current
X, Y, Z
(Talairach)
of
activated
voxels
X, Y, Z
(Talairach)
source
density
[AV/mm2
(�10�3)]
Motor Cortex �35, �27, 50 9435
Striatum left �25, �6, 6 4196
Striatum right 26, �5, 6 2642
Thalamus left �8, �16, 8 2881
Thalamus right 10, �12. 9 1782
Posteriorer
gyrus cinguli
0, �33, 44 142
Precuneus 9, �70, 35 171
Gyrus temporalis
inferior
left (y)
�52, �4, �27 3, 53
Gyrus temporalis
medialis
right (y)
3.53�13 4.19
Middle frontal
gyrus left
�45, 10, 29 3.59
(y) Region not covered by the fMRI slices.
C. Mulert et al. / NeuroImage 22 (2004) 83–9490
In the present study, LORETA-time-course peak detection
(Table 3) has been done by visual inspection.
Experimental procedure
Session in the scanner
Five runs with 120 scans each have been performed. During
each run, 15 target tones and 63 nontarget tones have been
presented, resulting in 75 target tones and 315 nontarget tones
for each subject. Between the end of the EPI sequence and the
stimulus presentation, there has been a quiet period of 1 s. The
Fig. 9. Current source density time courses. Left hemisp
EEG was recorded continuously during the entire experiment.
After stimulus onset, additional 580 ms poststimulus EEG could
be acquired until the next EPI sequence started.
Session in front of the scanner
Before the simultaneous EEG/fMRI measurement has been
performed, the same experiment has been done in front of the
scanner, using the same paradigm, the same number of stimuli, and
the same EEG equipment. Between the two runs, there was a break
of about 10 min.
In one subject, due to a technical problem of trigger marker
coding, no averaging was possible. All together, nine subjects
showed clear fMRI activations including the primary motor cortex
on a P value of 0.05, corrected for multiple testing, and were used
for further analyses.
Comparison between fMRI and EEG localizations
This comparison has been done in calculating the Euclidean
distances between the Talairach coordinates of the LORETA
current density maxima and the Talairach coordinates of the BOLD
centers of gravity in the same anatomical structures.
Results
Comparison between the EEG inside and in front of the scanner
There has been a clear difference in the amplitudes of the N1
component with higher amplitudes outside the scanner (mean:
10.15 F 5.70 AV vs. 5.06 F 2.58 AV, T = �3.37, P = 0.010).
However, no significant difference could be found between the
P300 potential with a mean amplitude at Pz of 8.00 F 4.25 AVoutside the scanner versus 6.60 F 3.98 AV inside, T = 1.39, P =
0.202 (Fig. 1). The N1 latency was longer outside the scanner
(mean: 118 F 10 vs. 109 F 9 ms, T = �3.130, P = 0.014). There
here (Talairach coordinates are given in Table 3).
Table 3
Time-course of the activations in the left and right hemisphere
Region P300 peak EEG (LORETA)
(ms)X, Y, Z (Talairach)
Left hemisphere
Gyrus temporalis inferior 294 � 52, � 4, � 27
STG 371 � 59, � 32, � 6
TPJ 385 � 59, � 25, 22
Lobulus parietalis inferior 387 � 45, � 46, 50
Insula 388 � 38, 17, 1
Middle frontal gyrus I 389 � 45, 10, 29
Middle frontal gyrus II 390 � 31, 38, 36
SMA/gyrus cinguli 396 � 3, � 11, 64
Precuneus 396 � 3, � 53, 57
Right hemisphere
Gyrus temporalis medialis 345 53, 3, � 13
Insula 356 53, 10, 8
TPJ 360 60, � 18, 15
Middle frontal gyrus 361 53, 17, 22
SMA/gyrus cinguli 396 4, � 11, 64
Precuneus 396 4, � 53, 57
Earlier activations in the right and the left insula, TPJ, and middle frontal
gyrus. SMA = supplementary motor cortex, TPJ = temporoparietal
Junction, STG = superior temporal gyrus.
C. Mulert et al. / NeuroImage 22 (2004) 83–94 91
was no significant difference between the P3 latencies outside and
inside the scanner (mean: 343 F 73 vs. 326 F 23 ms inside, T =
�0.63, P = 0.55). Both the frequent and the infrequent tone evoked
a P3a outside the scanner, but the P3a component of both the
frequent and infrequent tone inside the scanner was less distinct
(Fig. 2). In order to analyze the brain activity specifically related to
target detection, we have calculated difference waves (based on the
potentials related to the infrequent tone minus the potentials related
to frequent tone). Here, the specific target-related activity was
similar inside and outside the scanner (mean difference wave
amplitude at Pz inside the scanner 5.95 AV F 3.99 vs. 5.45 F3.49 outside the scanner, T = �0.35, P = 0.73). In addition, no
significant difference in the difference wave latencies was detected
(Fig. 3).
Comparison between fMRI and EEG localizations
fMRI activations were found in several brain regions including
the supplementary motor cortex (SMA), the anterior cingulate
cortex (ACC), the bilateral temporoparietal junction (TPJ), the
bilateral insula, and the frontal regions such as the middle frontal
gyrus (Figs. 4 and 5).
In the LORETA analysis, current density maxima have been
found in similar regions (Figs. 6, 7, and 8). Additionally, LOR-
ETA-maxima were found in the left inferior temporal gyrus (Z =
�27) and the right middle temporal gyrus (Z = �13). These
regions have not been covered by the fMRI slices. In the compar-
ison between the BOLD cluster centers of gravity and the LOR-
ETA-maxima, consistent spatial activation patterns were found
with Euclidean distances between 7 and 26 mm (mean of 16.0 F6.6 mm, Table 1).
Table 2 shows the regions with no correspondence between
fMRI and current source density including the motor cortex, the
subcortical regions (thalamus, striatum). Subcortical regions are
not included in the LORETA analysis, which is restricted on the
cortical gray matter and the hippocampus. Two fMRI activa-
tions, one in the posterior cingulate and one in the precuneus,
could not exactly be matched to a current source density
maximum. One current source density maximum in the left
Fig. 10. Current source density time courses. Right hemis
middle frontal gyrus did also have no direct equivalence in the
fMRI analysis.
Time-course analysis of the current source density maxima
Figs. 9 and 10 show the time course of brain activity in target
detection in the left and right hemispheres. In Table 3, the exact
peak latencies are given. There is early activity in the temporal
lobes, with the temporoparietal junction being active afterwards
and finally activation of frontal and supplementary motor regions.
Peak activity in frontal regions has been earlier on the right than on
the left, and also, activity in the insula was peaking earlier on the
right.
phere (Talairach coordinates are given in Table 3).
C. Mulert et al. / NeuroImage 22 (2004) 83–9492
Discussion
Comparison between the EEG inside the scanner and in front of
the scanner
Clear differences between the amplitudes of the N1 component
of the auditory-evoked potential inside the scanner and outside the
scanner could be detected, but no significant difference for the
P300 component. The N1 is a relatively early component and has
strong generators in the auditory cortex (Naatanen and Picton,
1987; Picton et al., 1999). The noisy environment inside the
scanner with the noise by the gradients has certainly an effect on
the activation of the auditory cortex: The gradient noise can thus be
seen as an additional auditory stimulus. In our study design,
between the quiet interval between the end of the scanner gradient
noise and the tone presentation was about 1 s. Therefore, the
auditory cortex has been in a partial refractory period and therefore
producing attenuated N1 amplitudes (Budd et al., 1998).
Outside the scanner, the same interstimulus interval of 3 or 6 s
has been used, but not interrupted by scanner (gradient) noise.
Therefore, both the infrequent and the frequent tone have evoked a
strong P3a component, representing an orientation reaction (Fried-
man et al., 2001). No such clear P3a component has been evoked
inside the scanner. With regard to the specific brain activity in
target detection, which has been analyzed in comparing the
difference waves of the potentials evoked by the infrequent minus
the potentials evoked by the frequent tones, the related activity was
quite similar inside and outside the scanner (Fig. 3). An effect of
the recording order of the two conditions is less likely since there
has been a break between the two runs and there is no learning
effect in an oddball paradigm.
Comparison between fMRI and EEG localizations
Beside the activity in the motor cortex, which is not stimulus-
locked and therefore not represented in the P300 potential, but
represented largely in the fMRI signal, all major cortical regions
have been well detected by the EEG-based LORETA analysis and
the fMRI analysis independently, including the temporoparietal
junction, the SMA/cingulate cortex, the insula, and the middle
frontal gyrus with a right > left asymmetry (Downar et al., 2000;
Halgren et al., 1998; Kiehl et al., 2001; Linden et al., 1999). Here,
an additional LORETA-maximum has been detected in the left
middle frontal gyrus with no correspondence in the fMRI signal.
Two additional LORETA-maxima in the left inferior temporal and
right middle temporal gyrus (Z = �13 and �27) have not been
covered by our fMRI slices (covering about Z = 55 to �10).
Finally, subcortical regions such as the thalamus and the striatum
have been strongly activated in the fMRI analysis. However,
thalamus and striatum are not included in the LORETA solution
space.
The mean Euclidean distance between EEG- and fMRI-based
localizations has been 16 mm, that is, in the range of the spatial
resolution of LORETA, which is between 1 and 2 cm. Our finding
is corresponding to a recent result of Vitacco et al. (2002)
describing a mean distance of 14.5 mm between fMRI and ERP
local maxima.
The results suggest that a close relationship exists between the
electrical P300-signal on the scalp and the corresponding BOLD
signal. Earlier studies, using from 19 to 124 electrodes (present
study: 27) and the LORETA-method or similar approaches, have
already described current source activity in parietal and frontal
regions, suggesting that the topographical information on the scalp
is sufficient for a plausible localization (Anderer et al., 2003;
Moores et al., 2003; Wang et al., 2003). The present study
demonstrates in addition a close relationship between EEG- and
fMRI-based localizations. This finding is in accordance with the
results of Horovitz et al. (2002) demonstrating a close correlation
between changes in the P300 amplitude and BOLD-signal changes
after manipulation of the experimental condition. The comparison
of the localizations in this study has been performed based on
current density images (ERP) and statistical images (fMRI). Beside
the good correspondence of the localizations, it may be best to use
similar methodology (e.g., the same kind of statistics) for the
different data types in the future, which has not been possible with
the applied software packages yet.
Time-course analysis of the current source density maxima
In both hemispheres, the activation sequence (within the P300
time range 250-540 ms) started in the temporal lobe with later
activations of the temporoparietal junction (TPJ) and latest activity
in frontal and supplementary cortex. Interestingly, the activity in
several brain regions (insula, TPJ, and SMA) has been peaking
earlier on the right than on the left side.
In comparison to the time-course patterns of the intracranial
recordings (Halgren et al., 1998), several differences are obvious:
In the present investigation, the time courses of the specific target-
related brain activity was analyzed, that is, the difference wave
between the potentials evoked by the infrequent versus frequent
tone. Furthermore, our paradigm did not evoke a strong early
(P3a)-component inside the scanner. Therefore, the time-course
patterns in our analysis do not show different time courses
contributing either to the P3a or P3b, but mainly different time-
course patterns within the P3b. Here, further studies are necessary,
using a paradigm that is evoking P3a and P3b inside the scanner
separately (Comerchero and Polich, 1999).
Applications in neuropsychiatric research
Concerning the possible clinical impact of this approach, we
suppose that a simultaneous measurement of EEG and fMRI,
enabling us to analyze data with high resolution spatial and
temporal information, should allow a distinct analysis of disturbed
brain function in diseases like schizophrenia, dementia, or depres-
sion. In these conditions, with very different pathologies, attenu-
ated P300 potentials can be recorded on the scalp (Polich and
Herbst, 2000).
Our finding of asymmetric time course patterns with earlier
activations in the TPJ, the insula, and the middle frontal gyrus on
the right in healthy subjects might be worth to be further inves-
tigated. Several authors are describing asymmetric P300 ampli-
tudes on the scalp in healthy subjects (Alexander et al., 1996;
Bruder et al., 1998). With a simultaneous EEG/fMRI approach,
this asymmetry could be further clarified, specifying the brain
regions involved and their respective time courses. Furthermore,
the relationship between P300 attenuations and psychopathology
could be further elucidated: Several authors have described re-
duced P300 amplitudes in patients with distinct thought disorder
(Frodl et al., 2002b; Higashima et al., 1998; Iwanami et al., 2000;
Juckel et al., 1996). Here, the combination of EEG and fMRI
might help to identify reduced activity in related brain regions and/
C. Mulert et al. / NeuroImage 22 (2004) 83–94 93
or a disturbed timing and synchrony of brain activity as a cerebral
correlate.
Based on the fact that high quality EEG recordings are possible
now in the scanner and that elimination of artifacts (originating,
e.g., from MR gradients) from the EEG-signal has recently much
improved (Allen et al., 1998; Goldman et al., 2000; Hoffmann et
al., 2000; Sijbers et al., 1999), further applications of a simulta-
neous EEG/fMRI approach are appropriate. For example, it is
obviously worthwhile to further investigate the relationship be-
tween electrical oscillations and the fMRI signal, since close
relationships have been described between the energy consuming
high frequency oscillations in the gamma (40 Hz)-range, and the
BOLD signal (Logothetis et al., 2001) and gamma oscillations are
also known to play a critical role in higher cognitive functions
(Engel et al., 2001). Furthermore, the investigation of EEG
coherence might improve our understanding of connectivity within
the brain, expanding our methodology here beyond the analysis of
hemodynamic-based connectivity approaches (Bokde et al., 2001;
Buchel and Friston, 2001).
In summary, this study shows diminished N1 amplitudes and
shorter N1 latencies inside the scanner but comparable P3 peaks
and latencies inside and outside the scanner as well as a high
degree of concordance between fMRI- and EEG-based localiza-
tions and distinct time-course patterns in the involved temporal,
parietal, and frontal regions.
Acknowledgment
Parts of this work were prepared in the context of Robert
Schmitt’s dissertation at the Faculty of Medicine, Ludwig-
Maximilians-University, Munich.
References
Alexander, J.E., Bauer, L.O., Kuperman, S., Morzorati, S., O’Connor, S.J.,
Rohrbaugh, J., Porjesz, B., Begleiter, H., Polich, J., 1996. Hemispheric
differences for P300 amplitude from an auditory oddball task. Int.
J. Psychophysiol. 21, 189–196.
Allen, P.J., Polizzi, G., Krakow, K., Fish, D.R., Lemieux, L., 1998. Iden-
tification of EEG events in the MR scanner: the problem of pulse
artifact and a method for its subtraction. Neuroimage 8, 229–239.
Anderer, P., Saletu, B., Semlitsch, H.V., Pascual-Marqui, R.D., 2003. Non-
invasive localization of P300 sources in normal aging and age-asso-
ciated memory impairment. Neurobiol. Aging 24, 463–479.
Baudena, P., Halgren, E., Heit, G., Clarke, J.M., 1995. Intracerebral poten-
tials to rare target and distractor auditory and visual stimuli: III. Frontal
cortex. Electroencephalogr. Clin. Neurophysiol. 94, 251–264.
Bokde, A.L., Tagamets, M.A., Friedman, R.B., Horwitz, B., 2001. Func-
tional interactions of the inferior frontal cortex during the processing of
words and word-like stimuli. Neuron 30, 609–617.
Bruder, G.E., Tenke, C.E., Towey, J.P., Leite, P., Fong, R., Stewart, J.E.,
McGrath, P.J., Quitkin, F.M., 1998. Brain ERPs of depressed patients to
complex tones in an oddball task: relation of reduced P3 asymmetry to
physical anhedonia. Psychophysiology 35, 54–63.
Buchel, C., Friston, K., 2001. Interactions among neuronal systems assessed
with functional neuroimaging. Rev. Neurol. (Paris) 157, 807–815.
Budd, T.W., Barry, R.J., Gordon, E., Rennie, C., Michie, P.T., 1998. Decre-
ment of the N1 auditory event-related potential with stimulus repetition:
habituation vs. refractoriness. Int. J. Psychophysiol. 31, 51–68.
Comerchero, M.D., Polich, J., 1999. P3a and P3b from typical auditory and
visual stimuli. Clin. Neurophysiol. 110, 24–30.
Downar, J., Crawley, A.P., Mikulis, D.J., Davis, K.D., 2000. A multimodal
cortical network for the detection of changes in the sensory environment
(In Process Citation). Nat. Neurosci. 3, 277–283.
Engel, A.K., Fries, P., Singer, W., 2001. Dynamic predictions: oscilla-
tions and synchrony in top-down processing. Nat. Rev., Neurosci. 2,
704–716.
Friedman, D., Cycowicz, Y.M., Gaeta, H., 2001. The novelty P3: an event-
related brain potential (ERP) sign of the brain’s evaluation of novelty.
Neurosci. Biobehav. Rev. 25, 355–373.
Frodl, T., Hampel, H., Juckel, G., Burger, K., Padberg, F., Engel, R.R.,
Moller, H.J., Hegerl, U., 2002a. Value of event-related P300 subcom-
ponents in the clinical diagnosis of mild cognitive impairment and
Alzheimer’s Disease. Psychophysiology 39, 175–181.
Frodl, T., Meisenzahl, E.M., Muller, D., Holder, J., Juckel, G., Moller, H.J.,
Hegerl, U., 2002b. P300 subcomponents and clinical symptoms in
schizophrenia. Int. J. Psychophysiol. 43, 237–246.
Gallinat, J.,Mulert, C., Bajobouji,M., Rentzsch, J., Senkowski, D., Schunter,
J., Mientus, S., Muchtieva, R., Kronfeldt, D., Nissle, S., et al., 2002.
Frontal and temporal dysfunction of auditory stimulus processing in
schizophrenia. Neuroimage 17, 110–127.
Goldman, R.I., Stern, J.M., Engel Jr., J., Cohen, M.S., 2000. Acquiring
simultaneous EEG and functional MRI. Clin. Neurophysiol. 111,
1974–1980.
Gray, C.M., Konig, P., Engel, A.K., Singer, W., 1989. Oscillatory responses
in cat visual cortex exhibit inter-columnar synchronization which
reflects global stimulus properties. Nature 338, 334–337.
Halgren, E., Baudena, P., Clarke, J.M., Heit, G., Liegeois, C., Chauvel, P.,
Musolino, A., 1995a. Intracerebral potentials to rare target and distrac-
tor auditory and visual stimuli. I. Superior temporal plane and parietal
lobe. Electroencephalogr. Clin. Neurophysiol. 94, 191–220.
Halgren, E., Baudena, P., Clarke, J.M., Heit, G., Marinkovic, K., Devaux,
B., Vignal, J.P., Biraben, A., 1995b. Intracerebral potentials to rare
target and distractor auditory and visual stimuli: II. Medial, lateral
and posterior temporal lobe. Electroencephalogr. Clin. Neurophysiol.
94, 229–250.
Halgren, E., Marinkovic, K., Chauvel, P., 1998. Generators of the late
cognitive potentials in auditory and visual oddball tasks. Electroence-
phalogr. Clin. Neurophysiol. 106, 156–164.
Hegerl, U., Juckel, G., Muller-Schubert, A., Pietzcker, A., Gaebel, W.,
1995. Schizophrenics with small P300: a subgroup with a neurodeve-
lopmental disturbance and a high risk for tardive dyskinesia? Acta
Psychiatr. Scand. 91, 120–125.
Higashima, M., Urata, K., Kawasaki, Y., Maeda, Y., Sakai, N., Mizukoshi,
C., Nagasawa, T., Kamiya, T., Yamaguchi, N., Koshino, Y., 1998. P300
and the thought disorder factor extracted by factor-analytic procedures
in schizophrenia. Biol. Psychiatry 44, 115–120.
Hoffmann, A., Jager, L., Werhahn, K.J., Jaschke, M., Noachtar, S., Reiser,
M., 2000. Electroencephalography during functional echo-planar imag-
ing: detection of epileptic spikes using post-processing methods. Magn.
Reson. Med. 44, 791–798.
Horovitz, S.G., Skudlarski, P., Gore, J.C., 2002. Correlations and disso-
ciations between BOLD signal and P300 amplitude in an auditory
oddball task: a parametric approach to combining fMRI and ERP.
Magn. Reson. Imaging 20, 319–325.
Horwitz, B., Poeppel, D., 2002. How can EEG/MEG and fMRI/PET data
be combined? Hum. Brain Mapp. 17, 1–3.
Iwanami, A., Okajima, Y., Kuwakado, D., Isono, H., Kasai, K., Hata, A.,
Nakagome, K., Fukuda, M., Kamijima, K., 2000. Event-related
potentials and thought disorder in schizophrenia. Schizophr. Res. 42,
187–191.
Jager, L., Werhahn, K.J., Hoffmann, A., Berthold, S., Scholz, V., Weber, J.,
Noachtar, S., Reiser, M., 2002. Focal epileptiform activity in the brain:
detection with spike-related functional MR imaging-preliminary results.
Radiology 223, 860–869.
Juckel, G., Muller-Schubert, A., Gaebel, W., Hegerl, U., 1996. Residual
symptoms and P300 in schizophrenic outpatients. Psychiatry Res. 65,
23–32.
C. Mulert et al. / NeuroImage 22 (2004) 83–9494
Kiehl, K.A., Liddle, P.F., 2001. An event-related functional magnetic reso-
nance imaging study of an auditory oddball task in schizophrenia. Schiz-
ophr. Res. 48, 159–171.
Kiehl, K.A., Laurens, K.R., Duty, T.L., Forster, B.B., Liddle, P.F., 2001.
Neural sources involved in auditory target detection and novelty pro-
cessing: an event-related fMRI study. Psychophysiology 38, 133–142.
Kiss, I., Dashieff, R.M., Lordeon, P., 1989. A parieto-occipital generator
for P300: evidence from human intracranial recordings. Int. J. Neurosci.
49, 133–139.
Linden, D.E., Prvulovic, D., Formisano, E., Vollinger, M., Zanella, F.E.,
Goebel, R., Dierks, T., 1999. The functional neuroanatomy of target
detection: an fMRI study of visual and auditory oddball tasks. Cereb.
Cortex 9, 815–823.
Llinas, R.R., 1988. The intrinsic electrophysiological properties of mam-
malian neurons: insights into central nervous system function. Science
242, 1654–1664.
Logothetis, N.K., Pauls, J., Augath, M., Trinath, T., Oeltermann, A., 2001.
Neurophysiological investigation of the basis of the fMRI signal. Na-
ture 412, 150–157.
Matsuda, T., Matsuura, M., Ohkubo, T., Ohkubo, H., Atsumi, Y., Tamaki,
M., Takahashi, K., Matsushima, E., Kojima, T., 2002. Influence of
arousal level for functional magnetic resonance imaging (fMRI) study:
simultaneous recording of fMRI and electroencephalogram. Psychiatry
Clin. Neurosci. 56, 289–290.
McCarthy, G., Wood, C.C., 1987. Intracranial recordings of endogenous
ERPs in humans. Electroencephalogr. Clin. Neurophysiol., Suppl. 39,
331–337.
Moores, K.A., Clark, C.R., Hadfield, J.L., Brown, G.C., Taylor, D.J.,
Fitzgibbon, S.P., Lewis, A.C., Weber, D.L., Greenblatt, R., 2003.
Investigating the generators of the scalp recorded visuo-verbal P300
using cortically constrained source localization. Hum. Brain Mapp. 18,
53–77.
Mulert, C., Gallinat, J., Pascual-Marqui, R., Dorn, H., Frick, K., Schlatt-
mann, P., Mientus, S., Herrmann, W.M., Winterer, G., 2001. Reduced
event-related current density in the anterior cingulate cortex in schizo-
phrenia. Neuroimage 13, 589–600.
Mulert, C., Juckel, G., Augustin, H., Hegerl, U., 2002. Comparison be-
tween the analysis of the loudness dependency of the auditory N1/P2
component with LORETA and dipole source analysis in the prediction
of treatment response to the selective serotonin reuptake inhibitor cit-
alopram in major depression. Clin. Neurophysiol. 113, 1566–1572.
Mulert, C., Gallinat, J., Dorn, H., Herrmann, W.M., Winterer, G., 2003. The
relationship between reaction time, error rate and anterior cingulate
cortex activity. Int. J. Psychophysiol. 47, 175–183.
Naatanen, R., Picton, T., 1987. The N1 wave of the human electric and
magnetic response to sound: a review and an analysis of the component
structure. Psychophysiology 24, 375–425.
Park, H.J., Kwon, J.S., Youn, T., Pae, J.S., Kim, J.J., Kim, M.S., Ha, K.S.,
2002. Statistical parametric mapping of LORETA using high density
EEG and individual MRI: application to mismatch negativities in
schizophrenia. Hum. Brain Mapp. 17, 168–178.
Pascual-Marqui, R.D., Michel, C.M., Lehmann, D., 1994. Low resolution
electromagnetic tomography: a new method for localizing electrical
activity in the brain. Int. J. Psychophysiol. 18, 49–65.
Pascual-Marqui, R.D., Lehmann, D., Koenig, T., Kochi, K., Merlo,
M.C., Hell, D., Koukkou, M., 1999. Low resolution brain electro-
magnetic tomography (LORETA) functional imaging in acute, neu-
roleptic-naive, first-episode, productive schizophrenia. Psychiatry Res.
90, 169–179.
Picton, T.W., Alain, C., Woods, D.L., John, M.S., Scherg, M., Valdes-Sosa,
P., Bosch-Bayard, J., Trujillo, N.J., 1999. Intracerebral sources of hu-
man auditory-evoked potentials. Audiol. Neuro-Otol. 4, 64–79.
Pizzagalli, D., Pascual-Marqui, R.D., Nitschke, J.B., Oakes, T.R., Lar-
son, C.L., Abercrombie, H.C., Schaefer, S.M., Koger, J.V., Benca,
R.M., Davidson, R.J., 2001. Anterior cingulate activity as a predictor
of degree of treatment response in major depression: evidence from brain
electrical tomography analysis. Am. J. Psychiatry 158, 405–415.
Polich, J., Herbst, K.L., 2000. P300 as a clinical assay: rationale, evalua-
tion, and findings. Int. J. Psychophysiol. 38, 3–19.
Polich, J., Pitzer, A., 1999. P300 and Alzheimer’s disease: oddball task
difficulty and modality effects. Electroencephalogr. Clin. Neurophy-
siol., Suppl. 50, 281–287.
Sijbers, J., Michiels, I., Verhoye, M., Van Audekerke, J., Van der Lin-
den, A., Van Dyck, D., 1999. Restoration of MR-induced artifacts in
simultaneously recorded MR/EEG data. Magn. Reson. Imaging 17,
1383–1391.
Silva, L.R., Amitai, Y., Connors, B.W., 1991. Intrinsic oscillations of neo-
cortex generated by layer 5 pyramidal neurons. Science 251, 432–435.
Soltani, M., Knight, R.T., 2000. Neural origins of the P300. Crit. Rev.
Neurobiol. 14, 199–224.
Talairach, J., Tournoux, P., 1988. Co-Planar Stereotaxic Atlas of the Human
Brain. Thieme, Stuttgart.
Towle, V.L., Bolanos, J., Suarez, D., Tan, K., Grzeszczuk, R., Levin, D.N.,
Cakmur, R., Frank, S.A., Spire, J.P., 1993. The spatial location of EEG
electrodes: locating the best-fitting sphere relative to cortical anatomy.
Electroencephalogr. Clin. Neurophysiol. 86, 1–6.
Vitacco, D., Brandeis, D., Pascual-Marqui, R., Martin, E., 2002. Corres-
pondence of event-related potential tomography and functional magnet-
ic resonance imaging during language processing. Hum. Brain Mapp.
17, 4–12.
Wang, J., Hiramatsu, K.I., Hokama, H., Miyazato, H., Ogura, C., 2003.
Abnormalities of auditory P300 cortical current density in patients with
schizophrenia using high density recording. Int. J. Psychophysiol. 47,
243–253.
Winterer, G., Mulert, C., Mientus, S., Gallinat, J., Schlattmann, P.,
Dorn, H., Herrmann, W.M., 2001. P300 and LORETA: comparison
of normal subjects and schizophrenic patients. Brain Topogr. 13,
299–313.