Posted: 13 January 1998

Spontaneous Slow and Fast MEG Activity in Schizophrenics
with Auditory Hallucinations

W. Sperling 1), M. Möller 2), H. Kober 2) J. Vieth 2), A. Barocka 1)
1)Dept. of Psychiatry and 2)Dept. of Experimental Neuropsychiatry, Friedrich-Alexander
University Erlangen-Nürnberg

Neurology, Psychiatry, and Brain Research, 4: 225-230, 1996


Hallucinations, perceptions in the absence of external stimuli are prominent among the cor symptoms of schizophrenia. The neural correlates of these brief, involuntary experiences are not well understood and have not been imaged selectively. A patient may experience hallucinations in more than one modality simultanously or at different times and they may or may not appear to emanate from a single source.
Auditory verbal hallucinations are thougt to arise from a disorder of inner speech (thinking in words), but little is known about how they arise. Recently neural correlates of tasks which involve inner speech have been examined in subjects with schizophrenia who hear voices by the use of rcBF, SPECT and PET.
In schizophrenic patients with hallucinations blood flow was significantly greater during hallucinations than in the non-hallucinating state in Broca`s area. Flow was also higher during hallucinations in the left anterior singulate cortex and regions in the left temporal lobe. The increased flow in Broca`s area was not accounted for by changes in other clinical variables nor by changes in the dose of neuroleptic drugs. Auditory hallucinations may also be reflected in distinctive metabolic maps of the brain. Regional brain metabolism was measured by positron emission tomography. Compared with the patients who did not experience hallucinations, the patients who did experience hallucinations had significantly lower relative metabolism in auditory and Wernicke`s regions and a trend towards higher metabolism in the striatum and anterior cingulate regions.
Neuroleptic treatment resulted in a significant increase in striatal metabolism and a reduced frontal -parietal ratio, which was significantly corrrelated with a decrease in hallucination scores. In serial assessments with 123-IMP SPECT an increased accumulation of 123-IMP in the left superior temporal area, which corresponds with the auditory association cortex was shown in schizophrenic patients with auditory hallucinations. So far, after several studies with different techniques in patients with auditory hallucinations the following areas seem to be regions of interest in the phenomenon of auditory hallucinations:

The new technique of Magnetoencephalography (MEG) could complete our knowledge about the brain topography of auditory hallucinations. MEG is a method of determining electrical activity in the brain noninvasively by detecting the magnetic fields associated with electric currents produced by neuronal activity. MEG has a very high time resolution, which can not be reached by other techniques.
Whereas scalp-EEG detects both tangential and radial sources, i.e. activity both in the sulci and in the gyri, MEG selectively measures tangential sources, i.e. activity in the sulci. Scalp EEG measures extracellular volume currents and MEG primarily intracellular currents. Finally, the spatial resolution of MEG is about 1/3 better compared to scalp-EEG (Vieth, 1992a, Vieth, 1992b). Besides that magnetic fields are not distorted by the different conductivities of the skull and the scalp as the EEG. With multichannel SQUID magnetometers, field patterns can be obtained and source localizations determined even with a "single shot" measurement without repositioning the instrument.


The 2x37-channel biomagnetic system (MAGNES II) of Biomagnetic Technologies, Inc. (BTi) has been used to record the MEG. The volume conductor (head) model, which we used, was spherical. The source model was the single dipole model. The MEG localization results have been inserted into an anatomical frame, obtained from the MRI scan. For this purpose it was necessary to determine the transformation parameters between the MRI and MEG-coordinate systems. Normally reference points were used at anatomical landmarks. But the number of these points was low, and the transformation accuracy was highly influenced by measuring errors. Therefore we developed a method to determine the transformation by a contour fit of the head surface. For our purpose an accessable and sufficient section of the head surface was scanned by an electromagnetic digitizer. Then the surface obtained by this procedure was fitted to the head surface reconstructed from the MRI scan. This transformation was exact (within 2 mmm), and was performed in more than 60 patients or subjects (Kober et al., 1992).

The 2x37 channel system allowed us to analyze spontaneous brain activity on both sides of the head with all its dynamics in space and time. The separation of sources was still one of the important questions and a challenge in brain neurophysiology. Magnetic field maps with two extrema of high intensitiy only were seen for short time sections changing in amplitude and gradually mixed with changing irregular patterns across time. When single dipoles were estimated this effect could still be seen, even when some precautions were taken to separate the sources.
We developed a method, which is able to show the concentrations of dipoles across time, the Dipole Densitiy Plot (DDP) (Kober et al., 1992, Vieth et al., 1992a, Vieth et al., 1992b). It is a spatial averaging in order to decrease the influence of the nonfocal activity. Noise is here the signal portion, which is not compatible with the single dipole model.

The DDP uses consecutively estimated dipoles across a given analyzing time and delivers quantified dipole concentrations in three dimensions, which can be adjusted exactly to individual slices of the imaging techniques. In order to minimize the influence of simultaneously active multiple sources, we applied two different procedures:

  1. Before the time consuming consecutive dipole estimation only those signal sections were automatically selected for the analysis, where the strongest component dominates the measured signal (typically at least 80 %), i.e., where predominantly only one source was active (Kober et al., 1992). This selection was done by using the principle component analysis (PCA). The procedure to find these sections was as follows: The correlation coefficient matrix of the field sensor was calculated across overlapping time intervals with a duration, which related to the upper frequency of the activity of interest. Typically the time section for theta/delta waves was 150 msec. Then a singular value decomposition of these correlation coefficient matrices was performed. The relative size of these singular values was a measure of the contribution of the different components of each component to the signal.
    Then only those time sections were used to localize single dipoles, where one component predominated the signal. Therefore the single dipole model was an adequate source model.

  2. In addition, after the single dipole estimation a selection of the dipoles was done on the basis of the S/N. Then only those dipoles were accepted for the further analysis, which were beyond a selected limit of this S/N (typically 2 to 3) The continuous version of the DDP determined the spatial distribution of dipoles by performing a three dimensional convolution by using a three dimensional Gaussian envelope, which took into account the localization uncertainty. The selected standard deviation (steepness of the envelope) was chosen according to the dipole localization error. The result was typically related to a volume of 1 cm3 , but nevertheless it was continously obtained in space. The quantified result were fused into the individual T1-MRI slices by using isocontour lines. The three dimensional plot could be shown with the number of dipoles in the Z-axis.

    In previous studies (Kober et al., 1992, Vieth et al., 1992a) results of our measurements in patients with structural lesions were all in close topographical relation to the infarctions, hemorrhages, tumors and cysts. There were also other approaches to show concentrations of dipoles across time, but none of them were that much elaborated as the DDP. We used the DDP-method now for spontaneous measurement in schizophrenic patients.


Spontaneous slow (2-6Hz) and fast MEG activity (12.5-30Hz) was measured in three schizophrenic patients (subtype:paranoid hallucinatory schizophrenia:ICD-10, F20.0). All patients had auditory hallucinations (imperative voices) during measurement. Two patients were treated with the neuroleptic haloperidole, one female patient was never treated with neurolpetics (first onset of illness). The control group consisted of three healthy probands. One day after her first measurement the female patient reported about no further auditory hallucinations. The MEG measurement in this patient was repeated on this day.


We separated the results of our measurement for Sensor A (left hemiphere) and Sensor B (right hemisphere). Table 1 shows the number, density and localisation of dipoles for slow (2-6Hz) and fast (12.5-30Hz) activity in schizophrenics and control probands for Sensor A, in table 2 for Sensor B. We also determined handedness (which is applied to be a peripheral laterality marker of cerebral dominance) in all probands (by the use of the Edingburgh handedness questionnaire).

Sensor A (left hemisphere):
In schizophrenic patients we found no striking effect for slow activity (2-6Hz) over the left hemisphere compared to control probands. In both groups slow activity was widely distributed, or concentrated on the central region of the left hemisphere.The dipole number and the dipole density, which means the percentage of one component, which dominates the signal (see methods), was even higher in healthy probands than schizophrenic patients (see table 1). A statistically significant difference was found in two schizophrenic patients for fast activity (12.5-30Hz) over the left hemisphere compared to control probands (see fig.1).
The standard density was about three times increased compared to the density values of healthy probands. One of these two schizophrenic patients was righthanded, the other ambidextrous. The localisation of fast activity dipoles in both cases was concentrated on the temporal region.

Sensor B (right hemisphere):
Dipole number and density for slow activity was increased in two control probands compared to schizophrenic patients (see table 2). In all probands (schizophrenic and healthy) slow activity dipoles were concentrated on the central region. For two schizophrenic patients we found a statistically significant increase for fast activity dipole density values. One patient was lefthanded (see fig. 4), the other was the same ambidextrous female patient, who showed a significant increase for fast activity dipoles over Sensor A.

Control MEG:
One day after her first MEG we measured the ambidextrous drug free female patient again, after she reported about a complete absence of auditory hallucinations. The standard density for fast activity dipoles was reduced from 10.6 to 4.5 (see table 1 and table 3) over the left hemisphere (see also fig. 6). Over the right hemisphere (see table 4) we still found a high fast activity in the second investigation. In the third investigation (three days later) the standard density for fast frequency was 3.8, which is a normal value.


Different symptoms of schizophrenia may be related to distinct types of brain dysfunction. It is possible to investigate the patterns of brain activity underlying schizophrenic symptoms by mapping regional metabolism or cerebral blood flow(rCBF) in vivo with positron emission tomography (PET) or single photon emission computed tomography (SPECT). Using these technique, some authors have related subsyndromes of symptoms to distinct rCBF patterns (Liddle et al.), wheras others have focused on the metabolic changes associated with more specific phenomena, most frequently auditory hallucinations (Cleghorn, MKcGuire.). Previous rCBF, PET and SPECT studies have assessed the patterns of activation in schizophrenic patients mainly in the temporal lobe. We used the new technique of magnetoencephalography to investigate schizophrenic patients with acute auditory hallucinations.
Primarily, if we had (without knowledge of our later results) to define hallucinations in terms of cerebral activation we would have preferred the model of cortical hyperactivity to that of hypoactivity. Finally we did not reject this hypothesis after our investigations. Slow activity dipoles were not increased in schizophrenic patients with auditory hallucinations over both hemispheres compared to healthy control probands. We rather had the impression of a certain increase of slow activity in healthy probands compared to schizophrenic patients. The statistically significant differences between both goups were found in the fast MEG activity of schozophrenic patients with auditory hallucinations. Opposite to the results of slow activity, which was localized in different regions of the hemispheres (central, frontal, parietal) we found a concentration effect for fast activity on the temporal region of the left and the right hemisphere. These results could be correlated to the peripheral laterality marker of handedness.
The significant increase of fast dipole density on the right hemisphere was found in a left handed patient, vice versa the increase on the left hemisphere in a right handed patient.

The schizophrenic ambidextrous patient had the increase of fast MEG activity in both hemispheres. If we accept the peripheral laterality sign handedness as a marker for cerebral dominance these results point to a stronger activation of the dominant hemisphere in auditory hallucinations. The principle of hemispheric dominance is supposed to be less strong in ambidextrous people. The increase of the dipole density for fast frequencies in the ambidextrous patient was in the left and the right hemisphere, a finding, which could confirm this hypothesis. In the same patient the clinical improvement with a complete absence of auditory hallucinations was correlated to a significant reduction of the density of fast activity over the left hemisphere, while the fast freuqency over the right hemiphere was not (not yet) altered. After a further investigation also the right hemisphere also reached normal values. Our MEG results confirm the findings of a strong participation ( in the sense of a hyperactivation) of the temporal lobe in auditory hallucinations.


  1. Busatto,G.F., David,A.S., Costa,D.C. (1995) Schizophrenic auditory hallucinations are associated with increased regional cerebral blood flow during verbal memory activation in a study using single photon emission computed tomography. Psychiatry Res. 61(4):255-264.
  2. Cleghorn,J.M., Franco,S., Szechtman,B. (1992) Toward a brain map of auditory hallcucinations. Am. J. Psychiatry 149(8): 1062-1069.
  3. Kober,H., Vieth,J., Grummich,P. (1992) The factor analysis used to improve the dipole density plot (DDP) to localize focal concentrations of spontaneous magnetic brain activity. Biomed. Engineer. (Berlin) 37(Suppl.2): 164-165.
  4. Liddle,P.F., Friston,K.F., Frithz,C.D. (1992) Patterns of cerebral blood flow in schizophrenia. Brit. J. Psychiatry 160: 179-186.
  5. McGuire,P.K., Shah,G.M., Murray,M. (1993) Increased blood flow in Broca's area during auditory hallucinations in schizophrenia. Lancet 342(8873): 692-693.
  6. Suzuki,M., Yusa,S., Minabe,Y. (1993) Left superior temporal blood flow increase in schizophrenic and schizophreniform patients with auditory hallucination: a longitudinal case study using 123-IMP SPECT. Eur.Arch. Psychiatry Clin. Neuroscien. 242(5): 257-261.
  7. Vieth, J., Kober, H., Sack, G., Grummich, P. (1992) The efficiacy of the discrete and the quantified continous dipole densitiy plot (DDP) in multichannel MEG. In: M.Hoke, Erne, S.N. (eds.) Biomagnetism Clinical aspects. Exerpta Medica. Amsterdam London, New York, 321-325.

Wolfgang Sperling, M.D.
Department of Psychiatry
Schwabachanlage 6-10
D-91054 Erlangen, Germany
Fax: +49-9131-855659
"Experimental Neuropsychiatry" Erlangen (deutsche Version)
Biomagnetism-Center Erlangen | Chapter "Biomagnetismus" of DGBMT