SEARCH

SEARCH BY CITATION

Keywords:

  • MEG;
  • Dynamic statistical parametric mapping;
  • Generalized spike;
  • Slow wave;
  • Epilepsy

Abstract

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. MEG/EEG, MRI, AND SPECT
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgments
  9. REFERENCES

Summary: Purpose: To examine whether magnetoencephalography (MEG) can be used to determine patterns of brain activity underlying widespread paroxysms of epilepsy patients, thereby extending the applicability of MEG to a larger population of epilepsy patients.

Methods: We studied two children with symptomatic localization-related epilepsy. Case 1 had widespread spikes in EEG with an operation scar from a resection of a brain tumor; Case 2 had hemispheric slow-wave activity in EEG with sensory auras. MEG was collected with a 204-channel helmet-shaped sensor array. Dynamic statistical parametric maps (dSPMs) were constructed to estimate the cortical distribution of interictal discharges for these patients. Equivalent current dipoles (ECDs) also were calculated for comparison with the results of dSPM.

Results: In case 1 with widespread spikes, dSPM presented the major activity at the vicinity of the operation scar in the left frontal lobe at the peak of the spikes, and some activities were detected in the left temporal lobe just before the peak in some spikes. In case 2 with hemispheric slow waves, the most active area was located in the left parietal lobe, and additional activity was seen at the ipsilateral temporal and frontal lobes in dSPM. The source estimates correlated well with the ictal manifestation and interictal single-photon emission computed tomography (SPECT) findings for this patient. In comparison with the results of ECDs, ECDs could not express a prior activity at the left temporal lobe in case 1 and did not model well the MEG data in case 2.

Conclusions: We suggest that by means of dSPM, MEG is useful for presurgical evaluation of patients, not only with localized epileptiform activity, but also with widespread spikes or slow waves, because it requires no selections of channels and no time-point selection.

The appropriate diagnosis of the epileptic syndrome based on the international classification of the epilepsies and epileptic syndromes is the most important factor in treatment decisions by the epileptologist (1). Currently, the diagnosis depends mainly on symptomatic and electrophysiologic findings. In this process, the analysis of interictal discharges (IIDs) and ictal discharges (IDs) in the EEG has played one of the most valuable parts. Magnetoencephalography (MEG) is complementary to EEG; the magnetic fields seen with MEG are selectively sensitive to tangentially oriented source current and are less influenced by the differences in conductivities in the head than are the electrical scalp potentials.

MEG recordings of IIDs can provide valuable information about the location of the epileptogenic area in the brain. Equivalent current dipole (ECD) analysis of MEG signals has suggested a good correspondence between IID and ID in determining the underlying epileptogenic area (2–4). The ECD model, however, is of limited use if the underlying assumption of focality is not fulfilled, as, for example, when the epileptiform activity occurs simultaneously across a wide area or when bilateral diffuse activity takes place, resembling generalized epileptiform discharges. For this reason, only those patients with symptomatic localization-related epilepsy (SLRE) with a focal epileptogenic area have been regarded as good candidates for MEG studies.

Recently Dale et al. (5) introduced the technique of dynamic statistical parametric mapping (dSPM) for the analysis of evoked MEG responses (5). In this approach, the cortical surface is partitioned into a large number of small patches, with each patch represented by an ECD in the middle of the patch, thus approximating any arbitrary spatial distribution of synaptic currents within the cortex. A further refinement of this method is achieved by normalizing the spatiotemporal estimates for noise sensitivity, thus yielding dSPMs of brain activity. By means of this technique, one can estimate the distribution of activity within the brain as a function of time.

This article presents an analysis of MEG recordings of paroxysms of epilepsy patients that appeared widespread from their initial detection and discusses whether the applicability of MEG can be extended to patients whose epileptic activity is not focal.

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. MEG/EEG, MRI, AND SPECT
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgments
  9. REFERENCES

Case 1

Case 1 was a 13-year-old girl who had seizures almost daily. At age 4 years, a primitive neuroectodermal tumor (PNET) was found at her left frontal lobe, associated with frequent vomiting, persistent fever, and loss of consciousness. Her tumor was totally resected at the same age, and she underwent radiation therapy (total of 54 Gy) and chemotherapy using the combination protocol with vincristine, cisplatin, and amino methyl pyrimidinyl methyl chlorethyl nitrosourea hydrochloride. At age 10 years, she began to have weekly seizures with impairment of consciousness. Although she had no seizures for 2 years under treatment with antiepileptic drugs (AEDs), her seizures reappeared, but with different semiology than before: complex gestural automatisms with total loss of consciousness, grunting, and incontinence. She could not recollect any scene during her seizures. Although she was treated by various kinds of AEDs [phenytoin (PHT), carbamazepine (CBZ), phenobarbital (PB), and clorazepate (CLP)], her seizures remained intractable. Her full-scale intelligence quotient (FIQ) assessed by Wechsler Intelligence Scale for Children, third version (WISC-III), was 60, verbal IQ was 62, and performance IQ was 66. She had no neurologic deficits.

Case 2

Case 2 was a 3-year-old boy who had started to have weekly seizures 4 months before the MEG examination. From age 1 year, he had febrile convulsive seizures twice. At age 3 years, he began to have afebrile seizures with clonic seizures of the right arm and twitching at the right lip with salivation, evolving to a right hemiconvulsive seizure. After a seizure, he usually had paralysis of the right arm and leg for ∼20 h. He complained of numbness in his right arm and leg just before his habitual seizure. Although he was treated with PHT, CBZ, and nitrazepam, his seizures did not cease. He grew normally. No interictal neurologic deficits were observed. His MRI was normal.

MEG/EEG, MRI, AND SPECT

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. MEG/EEG, MRI, AND SPECT
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgments
  9. REFERENCES

The two patients were studied at the Department of Pediatrics at Hokkaido University Hospital by using simultaneous recordings of MEG and EEG. The subjects' parents or guardians gave their written informed consent for the studies. MEG was recorded with a system with 204 superconducting quantum interference devices (SQUIDs) (Vectorview; Neuromag Ltd., Helsinki, Finland). This instrument has pairs of orthogonal planar gradiometers at each of 102 locations in a helmet-shaped array covering the entire scalp. This sensor configuration detects the maximal signal directly above the source current. The recordings were made in a magnetically and electrically shielded room. We recorded for duration of >1 h for each patient, collecting data in 4-min blocks. The raw MEG data were band-pass filtered between 0.03 and 133 Hz and sampled at 400 Hz. The patients were in a prone position with their heads inserted in the MEG dewar vessel. A sedative agent [pentobarbital (PTB)] was used because our patients were children. The relative positions of the head and the MEG sensors were determined by attaching three small head-position indicator coils to the head. The position of the coils was digitized and subsequently recorded by the MEG sensors for coregistration to MRI (6). EEG was recorded simultaneously for visual screening by using 20 scalp electrodes, placed according to the international 10–20 system, with two additional electrodes for electrocardiogram (ECG) monitoring. The EEG high-pass filtering time constant was 0.3 s.

Magnetic resonance images (MRIs) were acquired with a 1.5-T high-resolution MRI scanner (Magnetom VISION; Siemens AG, Erlangen, Germany) for diagnostic purposes and for supporting the analysis and interpretation of the MEG data (TE, 60 ms; TR, 100 ms; voxel size, 1.5 × 1.5 × 1.5 mm3).

For another diagnostic tool, the patients underwent single-photon emission computed tomography (SPECT) in the interictal period. We used a ring-type SPECT scanner (Headtome-SET070; Shimadzu Corp., Kyoto, Japan); [99mTc]-HMPAO was injected intravenously at a dose of 370 MBq in case 1 and 740 MBq in case 2.

MEG source analysis

The MEG data were digitally filtered with a band pass of 3 to 30 Hz for offline analysis. We manually selected the segments containing abnormal paroxysms. To determine the distribution of the brain activity generating the spikes, we used two source-estimation approaches: the equivalent current dipole model (ECD) and dynamic statistical parametric mapping (dSPM).

Equivalent current dipole model

We calculated ECDs by using the single-dipole model with dipole-fit software (Neuromag Ltd., Helsinki, Finland). The conductivity geometry of the head was assumed to be spherically symmetrical. Dipoles were calculated for each time sample within a period of 100 ms at the vicinity of each MEG spike, without deciding a region of interest (i.e., all sensors were included in the procedure). The ECD that had the best goodness of fit (GOF) was selected as the representative ECD of that particular MEG spike. GOF is a measure of how well the ECD model explains the measured signals. A dipole fit was acceptable with a GOF of >70% for case 1 and >40% for case 2. The ECDs were superimposed on each patient's MRI to visualize the anatomic location.

Dynamic statistical parametric mapping (dSPM)

The ECD is a good model when the underlying activity is focal (i.e., restricted to a relatively small region in the brain). For nonfocal activity, distributed source models are expected to be better suited than the ECD model (7). We applied an anatomically constrained linear estimation approach, which assumes the sources are distributed in the cerebral cortex (8). The cortical surface was reconstructed from the individual subjects' high-resolution T1-weighted MRIs (9–11). This surface was used for display of the estimated activity levels; it was inflated the better to visualize the sulcal cortex. The cortical surface also was subsampled to ∼2,500 elements per hemisphere, and the source model consisted of current dipole vectors located at each element (8). The magnetic field generated by each dipole component at each of these locations forward solution was calculated by using a boundary element model with conductivity boundaries determined from the segmented MRI (12,13). Only the inner surface of the skull is needed for MEG (13).

The time course of activity at each cortical location was estimate by using a noise-normalized linear estimation approach (5,14). This approach is similar to the generalized least-squares or weighted minimum norm solution (8,15), but the estimate is normalized for noise sensitivity, thus providing a statistical parametric map (5). The noise normalization reduces the variation in the point-spread function between locations (14). Simulations have suggested that the spatial resolution is 15 mm or better (5,14). Maps were calculated at 2.5-ms intervals. The significance of modulation at each site was calculated by using an F test (5,16). These statistical maps do not provide an estimate of the source strengths, particularly because the estimated noise variance is not constant across different cortical locations. Furthermore, source-strength estimates from MEG are subject to confounding influences of spatial extent of the sources and the variable amount of spatial cancellation. However, because the same noise covariance estimates are used at all time points for a given cortical location, source strength at a given location over time is directly proportional to the statistical maps. The current approach provides dynamic statistical parametric maps (dSPM) of cortical activity, similar to the statistical maps typically generated by using functional MRI (fMRI) or positron emission tomography (PET) data, but with a millisecond temporal resolution.

RESULTS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. MEG/EEG, MRI, AND SPECT
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgments
  9. REFERENCES

Case 1

Interictal EEG spikes frequently occurred dominantly at Fp1 and F3, but sometimes also at C3, F4, Fz, and F7 at the same time (Fig. 1A). Interictal MEG spikes were detected mainly at the left frontal MEG sensors, but additional activity appeared at the vicinity of the main activity at frontal through ipsilateral temporal and anterior midline sensors (Fig. 1B). ECDs for the interictal MEG spikes were located in the vicinity of the operation scar in the left frontal lobe. All ECDs shown had a GOF value of >70%. The majority of the ECDs were located at the lateral part of the scar; some ECDs, however, were located also at the medial part of the scar or mistakenly located at the vacant area in the operation scar (Fig. 2).

image

Figure 1. Interictal discharges in EEG and magnetoencephalography (MEG) (case 1). Top: Interictal discharge on EEG appeared dominantly at Fp1 and F3 with propagated spikes. Additional spikes at C3, F4, Fz, and F7 were sometimes found synchronously. Translucent yellow bar, The spiking points. Bottom: Interictal MEG spikes were detected at the left frontal region (surrounded by red line). However, the propagation of the spikes could be seen at the frontal region through the ipsilateral temporal region, as well as the anterior midline.

Download figure to PowerPoint

image

Figure 2. Source estimates for interictal magnetoencephalogram (MEG) spikes [equivalent current dipoles (ECDs, dynamic statistical parametric maps (dSPMs)] and interictal single-photon emission computed tomography (SPECT) (case 1). First column: representative paroxysms on MEG selected from the 204-channel SQUID signals. The two vertical lines indicate the time instants for the pre-interictal discharge (IID) and the peak of the IID for which the snapshots of ECDs and dSPM movies are shown (in the second, third, fourth, and fifth columns). Second column: ECD localization of Pre-IID for five different spikes. The locations of ECDs calculated for the IIDs are shown over a reconstructed cortical surface. The majority of the ECDs were located at the vicinity of operation scar at the left frontal lobe (green dots). The ECDs were selected by the value with >70% of goodness of fit (GOF). Most of the ECDs were located at the lateral part of the scar, some also were at the medial part. The red dots present the ECDs at the exact time points in each spike. In the first and second spikes, ECDs were located at the left frontal lobe at −56 ms and −68 ms, respectively (first and second row). Their GOF values were 82.2% and 80.0%, respectively. In the third, fourth, and fifth spikes, ECDs were located in or near the set of ECDs, and their GOF values were 76.0%, 66.2%, and 69.6%, respectively (third, fourth, and fifth row). Third column: ECD localization at the peak of spikes. Green dots present the majority of the ECDs fitted by the IIDs as second column. In the first, second, and third spikes, ECDs were located in the set of ECDs (first, second, and third row). Their GOF values were 83.2%, 77.6%, and 79.9%, respectively. In the fourth and fifth spikes, ECDs were located in the vacant area of the deficit of the brain at the peak of spikes, and their GOF values were 26.5% and 60.8%, respectively (fourth and fifth rows).Fourth column: Pre-IID dSPM maps for five different spikes. In the first and second spikes, weak activity at the tip of left temporal lobe appeared 56 and 68 ms before the peak of the spike. This was not seen in the maps for the other spikes. The dSPM maps are displayed on the inflated cortical surface, with darker gray indicating cortex buried in the sulci, and lighter gray indicating gyral cortex. The threshold of displayed activity is p < 0.001, and the full yellow indicates p < 10−9. Fifth column: dSPM maps for the peak of the IIDs. The most prominent activity occurred in the vicinity of the operation scar in the left frontal lobe (yellow). In the third spike, the maximum found in the statistical maps occurred at the inferior part of the operation scar, the lower bank of operculum of the left frontal lobe, and the tip of the temporal lobe. For the other spikes, the maximum was at the lateral part of the operation scar. Sixth column: Interictal SPECT ([99mTc]-HMPAO) indicating hypoperfusion at the vicinity of the operation scar, the basal ganglia, and the left anterior temporal lobe.

Download figure to PowerPoint

Several IIDs were analyzed with dSPM movies. These movies suggested that the whole area around the operation scar was active simultaneously when the spike occurred. Furthermore, minor ipsilateral temporal lobe excitation sometimes occurred 50–70 ms before the peak of a spike, although ECDs at the same time point were located at the frontal lobe instead of temporal lobe (Fig. 2).

An area of hypoperfusion, extending from the operation scar to the basal ganglia, was found in the left frontal lobe and the tip of the left temporal lobe in the interictal SPECT (Fig. 2).

Case 2

IIDs in the EEG were represented as nonlocalizing frequent lateralized left-hemispheric 2.5- to 3-Hz activity with a maximum in the central region (Fig. 3A). Left-hemispheric 7- to 10-Hz rhythmic discharges appeared just before and during his seizures. During convulsion, the recordings were contaminated by muscle artifacts. After the seizure, left-hemispheric slow activity appeared continuously.

image

Figure 3. Interictal discharges on EEG and magnetoencephalogram (MEG) (case 2). Top: The interictal discharges on EEG appeared as a left-hemispheric slow wave. Although these were most prominent in central electrodes, no obvious evidence of localized activity is seen. The translucent yellow bar presents the time points of slow waves. Bottom: Interictal discharges on MEG also were characterized as slow waves distributed over much of the left hemisphere, especially in posterior regions (surrounded by the red line).

Download figure to PowerPoint

Interictal discharges in MEG were described as intermittent slow waves over the left hemisphere that corresponded to 1- to 1.5-Hz slow waves on EEG (Fig. 3B). The locations of the ECDs fitted by these slow waves were widely scattered over both the parietal and the occipital lobes bilaterally. These ECDs were evaluated with wider statistic value: >40% of GOF. Movies of dSPM were made for three IID slow waves. These movies suggested a major excitation generated at the left supramarginal gyrus and postcentral gyrus in the left parietal lobe at the beginning of slow waves and propagated over the ipsilateral temporal and frontal lobe at or after the peak of slow waves. Instead of a focal region, the activated area at the peak of slow wave was already wide.

Hypoperfusion in the left parietal and frontal lobes was found in the interictal SPECT. SPECT images also were generated from an injection 15 min postictally, at a period when the right arm and leg showed Todd's paralysis. These images demonstrated a similar, but highly enhanced hypoperfusion (Fig. 4).

image

Figure 4. Equivalent current dipoles (ECDs), dynamic statistical parametric maps (dSPMs) for interictal MEG spikes and interictal SPECT (case 2). First column: ECDs fitted to interictal slow waves were localized over bilateral parietal lobes extending to occipital regions (green dots). No cluster was found among the ECDs. The ECDs were selected by the poor value with >40% of goodness of fit (GOF). Second column: Representative paroxysms on MEG. The vertical lines indicate when the dSPM snapshots were taken (shown in the third, fourth, and fifth columns). Third, fourth, and fifth columns: Selected time instants of the dSPM movies for three representative interictal discharges (IIDs). The maps suggest that the most active area (yellow) was in the left parietal lobe. In slow wave 1 (top), epileptic activity first occurred at the left supramarginal gyrus (−42 ms), gradually propagated to the left postcentral gyrus, the inferior and middle temporal gyrus, lateral occipital gyrus, and the lower bank of operculum in the inferior frontal gyrus at the peak (0 ms). Activity at the latter phase of this paroxysm occurred in the left lateral occipital gyrus (+34 ms). In slow wave 2 (middle), activity in the left post central gyrus (−16 ms) was accompanied by weaker activity at the left precentral gyrus and the left superior temporal gyrus at the peak (0 ms). Activity at the latter phase could be seen at the left pre- and postcentral gyrus and accompanied minor activity at the left middle frontal; gyrus (+34 ms). In slow wave 3 (bottom), the epileptic activity generated from the left postcentral and supramarginal gyrus (−16 ms) and propagated over insular cortex and anterior portion of the left temporal lobe at the peak (0 ms) (third row). At the latter phase of this paroxysm, activity was found in the left anterior superior temporal gyrus, insular cortex, and the operculum in the left inferior frontal gyrus (+34 ms). The threshold of displayed activity is p < 0.0007 for slow wave 1, and p < 0.005 for slow waves 2 and 3, with full yellow indicating p < 0.00005 for all. Sixth column: Interictal SPECT revealed hypoperfusion at the left parietal lobe and frontal lobe Seventh column: SPECT analysis when tracer was injected 15 min after his seizure, during temporary paralysis in his right arm and leg. The left parietal and frontal hypoperfusion was exaggerated in comparison with the purely interictal period.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. MEG/EEG, MRI, AND SPECT
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgments
  9. REFERENCES

The dSPM technique for MEG source analysis has been previously applied to studies of cognitive functions (5,16–19), illustrating dynamically changing brain function occurring simultaneously and sequentially in various cortical areas. In many studies of activity evoked by specific stimulation, multidipole analysis has been a good technique for modeling the source activity. The dipole model is based on the assumption that the brain activity generating the MEG/EEG signals consists of a small number of focal sources. In our study, however, because abnormal discharge seemed to occur simultaneously at widespread areas, the equivalent dipole analysis may not be able to characterize properly the active neuronal sources. Although ECDs to different IIDs located too many widespread areas in case 2, they would not be considered valid because they are not compatible with focal generators, as they had only low GOF value. However, the possibility still exists that the patient had multifocal epileptogenic area.

Our present study illustrates the clinical usefulness of dSPM in the analysis of widespread spikes or slow waves. In case 1, our results suggested a dynamically changing generator in a widespread prefrontal area surrounding the operation scar, as well as possible antecedent activity in anterior temporal cortex. The possibility of anterior temporal involvement was not proven by intracranial recordings but also was suggested by the interictal SPECT. In this patient, the calculated ECDs also were located around her scar. However, the validity of the ECD method in this case is questionable because the prior data suggested that the epileptogenic zone is widely distributed around the scar. Furthermore, the ECD method was not able to detect the antecedent temporal activity. In case 2, EEG was able to reveal only a left hemispheric slow wave, and the ECDs did not model the MEG data well. Although we tried to demonstrate some ECDs for case 2, those were inadequate examples of ECDs with only low GOF, and the localization of these ECDs was scattered over both hemispheres. The localization was clearly not a realistic solution, because the IIDs were limited to right hemisphere. In contrast, the dSPM movies revealed not only the main activity in the left parietal lobe but also demonstrated the propagation to other lobes. The ECD results in case 2 seemed to support a multifocal distribution of the epileptic foci. The dSPM movies, however, suggested that the irritative zone was located in the left parietal lobe. In this patient, his clinical seizure also corresponded well to our dSPM findings, as did the interictal SPECT. Thus in this case, dSPM appears to reveal the brain regions underlying the patient's symptoms more effectively than does EEG.

Previous researchers reported the propagation of the epileptiform activity by MEG with the result of ECDs calculated at each time point sequentially (20). They presented a patient with a focal epileptogenic focus from the temporal lobe. We suppose that sequential dipole fitting would not be acceptable in our case, because the described patient had widespread IIDs. We therefore conclude that dSPM analysis was a more reasonable estimation for our patients than the calculation of ECDs. In case 1, although both ECDs and dSPM techniques localized peak activity to similar locations, dSPM could see the earlier temporal lobe activity, which may be important to consider when planning surgical treatment or invasive presurgical diagnostic procedures or both. Additional activity in the temporal tip corresponded to the hypoperfusion in her interictal SPECT study. dSPM findings are of particular importance (beyond that of the SPECT) because the MEG gives the temporal information that the temporal lobe may act in some instances as a trigger for the prefrontal activity. Furthermore, her seizure consisted of complex gestural automatisms with complete loss of consciousness. It is relatively common that during frontal lobe seizures with gestural automatisms, patients retain consciousness and are able to recollect their circumstances at least during the beginning of the seizure (21). The lack of such awareness in this patient suggests prior activity, and thus is consistent with the possibility of antecedent temporal lobe activity suggested by the dSPM analysis. Although the single-dipole method can calculate an ECD at the time of discharge over the temporal tip, the amounts of GOF of this ECD were very low and evaluated as an insignificant result. In general, the solutions of ECD for MEG discharges are represented by the presentation of the major activity within the whole discharges, such as the peak of spikes. Therefore early epileptic propagation or preceding abnormal activities might have sometimes been neglected in the process of evaluation. dSPM can demonstrate propagation and preceding activities with high temporal resolution. In case 1, the epileptic discharges were not uniform. dSPM showed that some spikes were preceded by the ipsilateral temporal lobe, but some spikes did not have any preceding activities. Her polymorphic discharges might also imply the widespread epileptogenic area at the vicinity of the operation scar. Furthermore, some ECDs were calculated at clinically unrealistic places in this patient. These findings clearly indicate that this patient was not a good case for calculating ECD by using the single-dipole model. It might also be difficult to use the multidipole model, because it is difficult to select the channels for the analysis without any bias.

Compared with conventional ECD analysis, the dSPM method is based on a multifocal analysis, which calculates the activity in a semiautomated, unbiased way.

In case 2, the enhancement of the hypoperfusion in the SPECT obtained 15 min after the seizure, in comparison with interictal SPECT, could strongly suggest the existence of an irritative zone at the left parietal lobe (22). Using dSPM, we confirmed that the parietal lobe played a major role during the patient's IIDs and that the IIDs later propagated over another lobe. These findings also are supported by the prominent experiential and behavioral phenomena of his seizure semiology: feeling numbness within his right arm and easily evolving to right hemiconvulsive seizure and leaving Todd's palsy on the right arm. Slow waves are relatively nonspecific findings in EEG and do not always represent epileptic activities. However, especially intermittent rhythmic delta activity has been reported to be highly specific for diagnosing complex partial epilepsy (23), and some patients have only focal slow waves as their specific abnormal epileptiform discharges. For example, patients who had ischemic changes over the cerebral cortex were reported to demonstrate only slow waves as IIDs and IDs (24). Our findings further support the importance of epileptic slow waves in MEG that might also be interpreted for the evaluation of epileptic syndromes.

The use of dSPM as a new technique offers the possibilities of improving the presurgical evaluation in patients with medically intractable focal epilepsies, especially for patients with widespread IDs and IIDs in EEG and MEG. dSPM might help to optimize surgical resections and the position of invasive electrodes. This technique should also be valuable for an extended, deep evaluation of the nature of epileptic activity for patients who may not be candidates for epilepsy surgery, because uncovering the focal origin of a seemingly diffuse or generalized epileptic discharges may alter the choice of AEDs. The dSPM technique has been successfully applied to delineate a focal epileptogenic focus (25).

Although the data must be confirmed by electrocorticography (ECoG) as the gold standard, only a minor number of patients can be evaluated by such invasive monitoring. Because our patients were not thought to be good candidates for epilepsy surgery or invasive monitoring, we could not obtain confirmable findings of their epileptogenic foci.

It is clear that ictal information is the most valuable finding to diagnose the epileptic syndrome. Although no ictal events were found in our study, the dSPM technique might be applied also to ictal discharges recorded, because dSPM can express each activity at the seamless time points without setting a region of interest.

CONCLUSION

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. MEG/EEG, MRI, AND SPECT
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgments
  9. REFERENCES

These cases suggest that MEG may be a useful tool in the presurgical evaluation of patients with partial epilepsies, even when the epileptiform discharge appears to involve widespread regions simultaneously, provided that an appropriate distributed method is used for estimating the generating tissue. The use of dSPM can provide results without an operator-dependent bias that usually occurs with the selection of time points and sensor channels.

Acknowledgments

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. MEG/EEG, MRI, AND SPECT
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgments
  9. REFERENCES

Acknowledgment:  We thank Gregory P. Kirk, Maureen Glessner, Mark Halko, and Thomas Witzel for technical support. This study was supported by grants from MIND institute (DOE grant DE-FG03–99ER62764) and NIH (NS18741).

REFERENCES

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. MEG/EEG, MRI, AND SPECT
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgments
  9. REFERENCES
  • 1
    Commission on Classification and Terminology of the International League Against Epilepsy. Proposal for revised classification of epilepsies and epileptic syndromes. Epilepsia 1989;30: 38999.
  • 2
    Nakasato N, Levesque MF, Barth DS, et al. Comparisons of MEG, EEG, and ECoG source localization in neocortical partial epilepsy in humans. Electroencephalogr Clin Neurophysiol 1994;91: 1718.
  • 3
    Sutherling WW, Crandall PH, Engel JJ, et al. The magnetic field of complex partial seizure agrees with intracranial localizations. Ann Neurol 1987;21: 54858.
  • 4
    Shiraishi H, Watanabe Y, Watanabe M, et al. Interictal and ictal magnetoencephalographic study in patients with medial frontal lobe epilepsy. Epilepsia 2001;42: 87582.
  • 5
    Dale AM, Liu AK, Fischl BR, et al. Dynamic statistical parametric mapping: combining fMRI and MEG for high-resolution imaging of cortical activity. Neuron 2000;26: 5567.
  • 6
    Hamäläinen M, Hari R, Ilmoniemi RJ, et al. Magnetoencephalography: theory, instrumentation, and application to noninvasive studies of the working human brain. Rev Mod Phys 1993;65: 41397.
  • 7
    Dale AM, Halgren E. Spatiotemporal mapping of brain activity by integration of multiple imaging modalities. Curr Opin Neurobiol 2001;11: 2028.
  • 8
    Dale AM, Sereno MI. Improved localization of cortical activity by combining EEG and MEG with MRI cortical surface reconstruction: a linear approach. J Cognitive Neurosci 1993;5: 16276.
  • 9
    Dale AM, Fischl B, Sereno MI. Cortical surface-based analysis, I: segmentation and surface reconstruction. Neuroimage 1999;9: 17994.
  • 10
    Fischl B, Sereno MI, Dale AM. Cortical surface-based analysis, II: inflation, flattening, a surface-based coordinate system. Neuroimage 1999;9: 195207.
  • 11
    Fischl B, Liu A, Dale AM. Automated manifold surgery: constructing geometrically accurate and topologically correct models of the human cerebral cortex. IEEE T Med Imaging 2001;20: 7080.
  • 12
    Oostendorp TF, Van Oosterom A. Source parameter estimation in inhomogeneous volume conductors of arbitrary shape. IEEE T Biomed Eng 1989;36: 38291.
  • 13
    Hamäläinen MS, Sarvas J. Realistic conductivity geometry model of the human head for interpretation of neuromagnetic data. IEEE T Biomed Eng 1989;36: 16571.
  • 14
    Liu AK, Dale AM, Belliveau JW. Monte Carlo simulation studies of EEG and MEG localization accuracy. Hum Brain Mapp 2002;16: 4762.
  • 15
    Hamäläinen MS, Ilmoniemi RJ. Interpreting measured magnetic fields of the brain: estimates of current distribution. Helsinki : University of Technology, Dept. of Technical Physics Report TKK-F-A559, 1984.
  • 16
    Dhond RP, Buckner RL, Dale AM, et al. Spatiotemporal maps of brain activity underlying word generation and their modification during repetition priming. J Neurosci 2001;21: 356471.
  • 17
    Dhond RP, Marinkovic K, Dale AM, et al. Spatiotemporal maps of past-tense verb inflection. Neuroimage 2003;19: 91100.
  • 18
    Halgren E, Dhond RP, Christensen N, et al. N400-like MEG responses modulated by semantic context, word frequency, and lexical class in sentences. Neuroimage 2002;17: 110116.
  • 19
    Halgren E, Mendola J, Chong CDR, et al. Cortical activation to illusory shapes as measured with MEG. Neuroimage 2003;18: 10019.
  • 20
    Nakasato N, Kanno A, Shamoto H, et al. Magnetoencephalography in temporal lobe epilepsy: rapid propagation of early spike discharge to extratemporal areas. Epilepsia 2001;42: 534.
  • 21
    Penfield W, Jasper H. Epilepsy and functional anatomy of the human brain. Boston : Little, Brown, 1954: 51639.
  • 22
    Avery RA, Spencer SS, Spanaki MV, et al. Effect of injection time on postictal SPET perfusion changes in medically refractory epilepsy. Eur J Nucl Med 1999;26: 8306.
  • 23
    Normand MM, Wszolek ZK, Klass DW Temporal intermittent rhythmic delta activity in electroencephalograms. J Clin Neurophysiol 1995;12: 2804.
  • 24
    Kamada K. Saguer M, Moller M, et al. Functional and metabolic analysis of cerebral ischemia using magnetoencephalography and proton magnetic resonance spectroscopy. Ann Neurol 1997;42: 55463.
  • 25
    Shiraishi H, Stufflebeam SM, Knake S, et al. Dynamic statistical parametric mapping (dSPM) for analyzing the magnetoencephalographic epileptiform activity in patients with epilepsy. J Child Neurol 2005;20: 3639.