Elsevier

NeuroImage

Volume 22, Issue 1, May 2004, Pages 83-94
NeuroImage

Integration of fMRI and simultaneous EEG: towards a comprehensive understanding of localization and time-course of brain activity in target detection

https://doi.org/10.1016/j.neuroimage.2003.10.051Get rights and content

Abstract

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 Ā± 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 electromagnetic 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 Ā± 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.

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 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, Halgren et al., 1995b.

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 noninvasive 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 in target detection became possible now with high spatial resolution (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 hemodynamic-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 studies in monkeys (Logothetis et al., 2001), investigating simultaneously 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 experimental 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 implications 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 simultaneously with both methods. Thus, to get usable and interpretable results, it is worthwhile to make more effort concerning methodology. 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?

Section snippets

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, 20ā€“30 years old; mean age, 24.2 Ā± 2.9) were analyzed. The study was approved by the local ethics committee of the Ludwig-Maximilians-University of Munich and written informed consent was obtained from each subject.

Paradigm

Auditory

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 Ā± 5.70 Ī¼V vs. 5.06 Ā± 2.58 Ī¼V, T = āˆ’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 Ā± 4.25 Ī¼V outside the scanner versus 6.60 Ā± 3.98 Ī¼V inside, T = 1.39, P = 0.202 (Fig. 1). The N1 latency was longer outside the scanner (mean: 118 Ā± 10 vs. 109 Ā± 9 ms, T = āˆ’3.130, P = 0.014). There was no

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

Acknowledgements

Parts of this work were prepared in the context of Robert Schmitt's dissertation at the Faculty of Medicine, Ludwig-Maximilians-University, Munich.

References (56)

  • R.I Goldman et al.

    Acquiring simultaneous EEG and functional MRI

    Clin. Neurophysiol.

    (2000)
  • E Halgren et al.

    Intracerebral potentials to rare target and distractor auditory and visual stimuli. I. Superior temporal plane and parietal lobe

    Electroencephalogr. Clin. Neurophysiol.

    (1995)
  • E Halgren et al.

    Intracerebral potentials to rare target and distractor auditory and visual stimuli: II. Medial, lateral and posterior temporal lobe

    Electroencephalogr. Clin. Neurophysiol.

    (1995)
  • E Halgren et al.

    Generators of the late cognitive potentials in auditory and visual oddball tasks

    Electroencephalogr. Clin. Neurophysiol.

    (1998)
  • M Higashima et al.

    P300 and the thought disorder factor extracted by factor-analytic procedures in schizophrenia

    Biol. Psychiatry

    (1998)
  • S.G Horovitz et al.

    Correlations and dissociations between BOLD signal and P300 amplitude in an auditory oddball task: a parametric approach to combining fMRI and ERP

    Magn. Reson. Imaging

    (2002)
  • A Iwanami et al.

    Event-related potentials and thought disorder in schizophrenia

    Schizophr. Res.

    (2000)
  • G Juckel et al.

    Residual symptoms and P300 in schizophrenic outpatients

    Psychiatry Res.

    (1996)
  • K.A Kiehl et al.

    An event-related functional magnetic resonance imaging study of an auditory oddball task in schizophrenia

    Schizophr. Res.

    (2001)
  • C Mulert et al.

    Reduced event-related current density in the anterior cingulate cortex in schizophrenia

    Neuroimage

    (2001)
  • C Mulert et al.

    Comparison between 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 citalopram in major depression

    Clin. Neurophysiol.

    (2002)
  • C Mulert et al.

    The relationship between reaction time, error rate and anterior cingulate cortex activity

    Int. J. Psychophysiol.

    (2003)
  • R.D Pascual-Marqui et al.

    Low resolution electromagnetic tomography: a new method for localizing electrical activity in the brain

    Int. J. Psychophysiol.

    (1994)
  • R.D Pascual-Marqui et al.

    Low resolution brain electromagnetic tomography (LORETA) functional imaging in acute, neuroleptic-naive, first-episode, productive schizophrenia

    Psychiatry Res.

    (1999)
  • J Polich et al.

    P300 as a clinical assay: rationale, evaluation, and findings

    Int. J. Psychophysiol.

    (2000)
  • J Sijbers et al.

    Restoration of MR-induced artifacts in simultaneously recorded MR/EEG data

    Magn. Reson. Imaging

    (1999)
  • V.L Towle et al.

    The spatial location of EEG electrodes: locating the best-fitting sphere relative to cortical anatomy

    Electroencephalogr. Clin. Neurophysiol.

    (1993)
  • J Wang et al.

    Abnormalities of auditory P300 cortical current density in patients with schizophrenia using high density recording

    Int. J. Psychophysiol.

    (2003)
  • Cited by (518)

    View all citing articles on Scopus
    View full text