Dipole Localization for Identification of Neuronal Generators in Independent Neighboring Interictal EEG Spike Foci

Transcription

1 Epilepsia, 42(4): , 2001 Blackwell Science, Inc. International League Against Epilepsy Clinical Research Dipole Localization for Identification of Neuronal Generators in Independent Neighboring Interictal EEG Spike Foci *, Ayako Ochi, *, Hiroshi Otsubo, *, Shiro Chitoku, Amrita Hunjan, Rohit Sharma, James T. Rutka, Sylvester H. Chuang, Ken-ichi Kamijo, Toshimasa Yamazaki, and *, O. Carter Snead III *Bloorview Epilepsy Research Program, and Division of Neurology, Department of Paediatrics, Division of Neurosurgery, Department of Surgery, and Department of Diagnostic Imaging, The Hospital for Sick Children and the University of Toronto, Toronto, Ontario, Canada; and Fundamental Research Laboratories, NEC Corporation, Tsukuba, Japan Summary: Purpose: We evaluated dipole localizations of independent neighboring interictal spike foci using scalp electroencephalogram (EEG) to identify neuronal generators of epileptic discharges. Methods: Three pediatric patients with extratemporal lobe epilepsy who had two independent neighboring interictal spike foci on scalp EEG were studied. Prolonged video EEG was digitally recorded from 19 scalp electrodes, whose positions were registered using a three-dimensional digitizer. Interictal spikes were visually selected based on negative phase reversals on bipolar montages. We analyzed the dipole position and moment of each spike using a single moving dipole and three-shell spherical head model. The dipoles were overlaid onto magnetic resonance (MR) images and divided into two groups based on two spike foci. Results: The dipoles of the two groups were oriented either tangentially or radially to the scalp in close proximity to each other. The dipoles oriented radially were located underneath the electrode with a negative peak; those oriented tangentially were between electrodes with a negative and positive peak. The positions of tangential dipoles were more concentrated than those of radial dipoles. The epileptogenic regions corresponded to the dipole localizations. Surgical excisions were performed based on the results of electrocorticography. After surgery, two patients were seizure free, and one had rare seizures (follow-up period, months). Conclusions: We showed that dipoles in close proximity but with different orientations projected two negative maxima on scalp EEG in three patients with extratemporal localizationrelated epilepsy. Equivalent current dipole analysis of individual interictal spikes can provide useful information about the epileptogenic zone in these patients. Key Words: Tangential dipole Phase reversal EEG Dipole orientation Children. The main tasks in routine electroencephalography (EEG) are to recognize waveforms of diagnostic significance and identify the location of their generators within the brain (1). Studies (1 3) based on the solid angle theorem of volume conductor theory have shown that a higher amplitude does not necessarily mean that the recording electrodes are closer to the neuronal source, because the generators of the EEG are cortical pyramidal neurons, which have dipolar configurations. Theoretically, in the EEG, when the polarity anywhere on the scalp is either negative or positive, such a field Accepted December 11, Address correspondence and reprint requests to Dr. A. Ochi at Division of Neurology, The Hospital for Sick Children, 555 University Avenue, Toronto, Ontario, M5G 1X8, Canada @ ican.net would resemble that generated by a dipole layer located in the crown of the gyrus, often called a radial dipole. If the generator occupies a fissural cortex, the potential distribution on the scalp shows both positive and negative polarities. Such a field is often referred to as a tangential dipole. Through variations in orientation and shape of the cerebral cortices, dipole layers can produce charge reorientation and cancellation (4). Equivalent current dipoles should be considered the best representation of the center of a dipole layer that corresponds to the activity of a patch of cortex (5). Previous studies of dipole localizations of interictal EEG spikes have considered the differences among dipole orientations (6 9). Benign rolandic epilepsy of childhood without neurologic abnormalities (the typical form) demonstrated a stable tangential dipole, whereas 483

2 484 A. OCHI ET AL. more than one dipole source was observed in patients with the atypical form of the disorder, which includes neurologic or intellectual abnormalities (6 8). Van der Meij et al. (9) described the double spike phenomenon, consisting of a rolandic spike preceded by a small spike in its ascending phase. These two spike sources derived from averaged spikes were located close together but showed a different orientation. In the current study, we hypothesized that a single epileptic region could produce two independent neighboring spike foci on a scalp EEG. In three children with refractory extratemporal lobe epilepsy, we describe the relationship between (a) positions and orientations of two dipole groups using individual spikes, and (b) the electrode positions where the maximal voltage was recorded. METHODS Patients Five patients with extratemporal lobe epilepsy who were studied for dipole localization underwent epilepsy surgery at The Hospital for Sick Children between 1996 and Three of them had two independent neighboring interictal spike foci noted on EEG, and were selected for study. Their parents gave informed consent for their participation in the study. Data acquisition Digital prolonged video-eegs were recorded from 19 scalp electrodes placed according to the International system (BMSI 5000; Nicolet, Madison, WI, U.S.A.). A single reference was used at FCz in patient 1 and Oz in patients 2 and 3. The sampling rate was 200 Hz (a 1- to 70-Hz bandpass filter). Interictal spikes were visually selected and categorized according to the electrode of negative phase reversal on either longitudinal or transverse bipolar montages. The scalp electrode positions were registered using 3SPACE ISOTRAK II (Polhemus, Colchester, VT, U.S.A.) combined with equivalent current dipole localization software for Windows, SynaPointPro (GE Marquette Medical System Japan, Ltd., Tokyo, Japan). After the electrode positioning had been registered, measured electrode locations were fitted to the sphere (10). In addition, three fiduciary points (nasion, and right and left preauricular points) were obtained for EEG and magnetic resonance imaging (MRI) coregistration. Figure 1 shows the coordinate system used in SynaPointPro. Data analysis EEG spike samples were filtered with a high pass of 5 Hz and a low pass of 35 Hz, with a notch filter (60 Hz) for equivalent current dipole analysis. We analyzed the dipole localization of EEG spikes using a single moving dipole inverse-solution algorithm, a three-shell spherical FIG. 1. Coordinate system for EEG dipole position and moment in the SynaPointPro program. The center of the head is defined as the point where the y-axis (nasion inion line) perpendicularly intersects the x-axis (auricular line, with positive value right of center and negative value left of center). The z-axis extends from the center toward the vertex of the head at a right angle. Values for the center of the head (x, y, z) are all zero. A closed circle shows a dipole position, and a tail represents a dipole moment, which includes orientation and strength of an equivalent current dipole. Direction of the tail indicates positivity. head model, and electrode-positioning data (SynaPoint- Pro). We have previously reported our algorithm and methods of equivalent current dipole analysis (10,11). In the current study, goodness of fit (GoF) was applied instead of residual variance [GoF 100 residual variance (%)]. The algorithm of this program calculates the inverse solution from 10 random, computer-generated initial guesses within the head model and chooses the highest GoF at each time point because the inverse problem has no unique solution. Duration of 100 ms before and after the maximal negative peak was examined in each spike. The dipole fits occurred every 5 ms. We chose the dipoles that showed a GoF >95%and that existed from the spike s onset to its maximal negative peak. We calculated the angle between the mean dipole orientations of two independent spike groups, using the cosine of each normalized moment. Neuroimaging The dipole localizations were overlaid onto the MR images of each patient s brain. MRI (GE Signa 1.5 Tesla; General Electric Medical Systems, Milwaukee, WI, U.S.A.) yielded continuous 124 T 1 -weighted coronal slices with a thickness of 2 mm, a pixel size of mm, and a image matrix. The SynaPointPro program was used for superimposition of the dipoles on the MR images (coronal, axial, and sagittal views). RESULTS Table 1 shows the clinical profiles of the three patients studied. The total number of spikes and time points analyzed for dipole localizations and selected by our criteria, according to the electrode with a negative phase reversal,

3 DIPOLE IN INDEPENDENT NEIGHBORING SPIKES 485 Patient Age (y) TABLE 1. Clinical profiles of three patients MRI findings Electrocorticography Dipole Surgical procedures Pathological findings Outcome (follow-up period) 1 6 L P cyst ECoG/L P L P Lesionectomy + MST Porencephalic cyst Sz-free (2 yr, 3 mo) 2 15 Normal IVEEG/R F, T, P R F R F, T, cortical excision + P, MST Gliosis Sz-free (2 yr, 7 mo) 3 13 Normal IVEEG/R F R F R F, cortical excision + MST Gliosis Rare Sz (13 mo) L, left; R, right; P, parietal lobe; F, frontal lobe; T, temporal lobe; ECoG, intraoperative electrocorticography; IVEEG, intracranial invasive video EEG monitoring; MST, multiple subpial transection; Sz, seizure; mo, months. are shown in Table 2. Table 3 presents the mean, standard deviation (SD), and range of the dipole positions, the mean of the dipole orientations, and the angle between the mean dipole orientations, according to the electrode with a negative phase reversal in each patient. Patient 1 In patient 1, scalp EEGs showed two independent negative phase reversals at electrodes Pz and C3 in a transverse bipolar montage (Fig. 2). The spikes with a negative phase reversal at Pz (Pz group) had a positive phase reversal at C3 simultaneously. The dipoles of the Pz group were located between electrodes Pz and C3 and oriented tangential to the scalp. The dipoles of the spikes with a negative phase reversal at C3 (C3 group) were close to the electrode C3 and oriented radially (Fig. 2). Figure 2 (right) illustrates the distance between the electrodes and the dipoles. The SD of the dipole positions from the Pz group was smaller than that from the C3 group (Table 3). The dipoles of the Pz group were located on the superior medial walls of the left parietal cyst, and those of the C3 group on the inferior lateral side of the wall on the coregistered MR image. In three of 17 Pz spikes, a negative maximum at Pz shifted to C3 15 ms after the Pz peak: these dipole positions and orientations during propagation were identical to those in the Pz to C3 group. Patient 2 In patient 2, the longitudinal bipolar montage EEG showed independent negative phase reversals at F8 and F4 (Fig. 3). The spikes with a negative phase reversal at F8 (F8 group) had a small positive polarity at Cz and Pz simultaneously on the voltage topography, but this was not clear on the bipolar montage EEG (Fig. 3, color contour map). The voltage topography of the spikes with a negative phase reversal at F4 (F4 group) showed a single negative field. The mean dipole position of the F8 group was remote from electrodes F8 and Cz, respectively, whereas that of the F4 group was close to electrode F4 (Fig. 3, right). The distance between the mean dipole positions of each group was much shorter than that between electrodes F8 and F4. There was no propagation of dipoles between the two groups. The dipoles in each group were stable and independent. The dipoles of both F8 and F4 groups were localized in the right frontal region on the coregistered MR image. Patient 3 In patient 3, the longitudinal bipolar montage EEG showed independent negative phase reversals at F8 and F4 (Fig. 4). The spikes with a negative phase reversal at F8 (F8 group) were accompanied by a positive phase reversal at F3. The voltage topography of the F8 group showed a tangential dipole pattern with a negative polarity at right frontal region and a positive polarity at left frontal region, whereas that of the spikes with a negative phase reversal at F4 (F4 group) showed only a negative polarity at right frontal region (Fig. 4, color contour map). Five of 17 F4 spikes had a small positivity around electrodes P4 or T6 on the voltage topography, in which dipoles were located in the posterior frontal region (Fig. 4, bottom axial MR image). The dipole of the F8 group was located between electrodes F3 and F8 and oriented tangential to the scalp, whereas that from F4 group was close to the electrode F4 and oriented radially (Fig. 5). The dipoles were located close to each other. The range and SD of the dipole position in the F8 group were Patient Total analyzed spikes TABLE 2. Number of analyzed spikes and time points Electrode at negative phase reversal Number of spikes Number of spikes with GoF > 95% (%) Number of analyzed time points Number of dipoles with GoF > 95% (%) 1 37 Pz (100) (6.7) C (100) (5.7) 2 62 F (38.9) (1.8) F (34.6) (1.8) 3 69 F (60.9) (2.7) F (37.0) (1.8) GoF, goodness of fit.

4 486 A. OCHI ET AL. Patient Dipoles TABLE 3. Mean dipole position and orientation Mean dipole position (mm) ± SD (range) Mean dipole orientation x a y z x y z Angle between groups b 1 Pz group 27.5 ± ± ± (n 47) ( ) ( ) ( ) C3 group 55.1 ± ± ± (n 47) ( ) ( ) ( ) 2 F8 group 11.7 ± ± ± (n 27) ( ) ( ) ( ) F4 group 11.5 ± ± ± (n 19) ( ) ( ) ( ) 3 F8 group 24.2 ± ± ± (n 25) ( ) ( ) ( ) F4 group 27.8 ± ± ± (n 34) ( ) ( ) ( ) SD, standard deviation. a Values of x, y, and z are coordinate system for dipole position and orientation (Fig. 1). b The angle between the mean dipole orientations of the two groups. smaller than those in the F4 group (Table 3). There was no propagation of the dipoles between the groups. The dipoles of the F8 group were concentrated in the right anterior frontal region, whereas those of the F4 group showed diffuse distribution from the right anterior to the posterior frontal region on MR imaging (Fig. 4). Tangential and radial dipoles In all three patients, the dipoles of two independent neighboring interictal spike foci from the scalp EEG showed both tangential and radial orientations. The radial dipoles were located close to the electrode at the negative phase reversal, and the tangential dipoles be- FIG. 2. The results of dipoles from two interictal EEG spikes in patient 1. At left, two types of interictal spikes, one with a negative phase reversal at Pz (Pz group) and the other with a negative phase reversal at C3 (C3 group) are shown. Black dots indicate negative phase reversals in a transverse bipolar montage. Pz group had a positive phase reversal at C3 (open square). Center, axial head models show the dipoles of Pz group (top, small yellow circles and black tails) and those of C3 group (bottom, small red circles and black tails). Yellow and red circles show electrode positions of Pz and C3, respectively, and blue circles show the other electrodes. At right, a schema illustrates the distance (millimeters) between the dipoles, electrodes, and a left parietal cyst.

5 DIPOLE IN INDEPENDENT NEIGHBORING SPIKES 487 FIG. 3. The results of dipoles from two interictal EEG spikes in patient 2. At left, two types of interictal spikes, one with a negative phase reversal (black dots) at F8 (F8 group), and the other with a negative phase reversal at F4 (F4 group), are shown in a longitudinal bipolar montage. Center, the voltage topographies at the time point with a negative maximum of each spike show a tangential dipole pattern in F8 group (top) and radial dipole pattern in F4 group (bottom). At right, a schema represents the distance (millimeters) between dipoles (F8 group, small yellow circle and black tail; F4 group, small red circle and black tail) and electrodes (yellow, F8; red, F4; green, Cz) in the right frontal region. tween the electrode at the negative and positive phase reversals, all located close together. The tangential dipole positions were more concentrated than the radial. Surgical excisions were performed based on the results of either intraoperative electrocorticography or intracranial invasive video-eeg monitoring (Table 1). The epileptogenic regions corresponded to the dipole localizations on EEG. After surgery two patients (1 and 2) were seizure free, and one patient (3) had seizures only rarely (follow-up period, months). DISCUSSION We analyzed equivalent current dipoles from interictal spikes in children whose EEGs showed two independent neighboring spike foci. The dipole positions of these foci were in close proximity, but because of the difference in their orientation, two different negative potential fields were projected on the scalp EEG. Although equivalent current dipole analysis oversimplifies the brain s complicated neuronal activities, we were able to delineate some aspects of the relationship between neuronal generators and scalp potential fields, using actual values of dipole parameters, electrode positions, and coregistered MR images. We demonstrated that the radial dipoles were located close to the electrode at the negative maximum of the spike, and the tangential dipoles between negative and positive maxima. A potential difference recorded between two electrodes depends more on the orientation of the generator than on its proximity to the electrodes. If a generator of synchronized activity occupies only the crown of a gyrus on the convexity of the brain, the largest potential would be recorded by an electrode facing the midportion of the crown of the gyrus, and the potential profile along a line on the scalp shows a bell-shaped curve (1) (Fig. 5, bottom, solid line). Conversely, if the generator occupies a fissural cortex oriented orthogonal to the scalp surface, an electrode directly above it is on the zero isopotential line and records no voltage (1,12). Potential distribution along a line on the scalp shows an S-shaped curve (1) (Fig. 5, bottom, dotted line). Our study confirmed the solid-angle concept presented by Gloor (1), using equivalent current dipole analysis, the actual location of electrodes measured by threedimensional digitizer, and coregistered MR images. Our results demonstrated that the distance between tangential and radial dipoles was smaller than that between independent negative maxima as shown on scalp EEG. In previous studies (9,13,14), dipoles located in the

6 488 A. OCHI ET AL. FIG. 4. The results of dipoles from two interictal EEG spikes in patient 3. At left, two types of interictal spikes, one with a negative phase reversal (black dots) at F8 (F8 group) and the other with a negative phase reversal at F4 (F4 group) are shown in a longitudinal bipolar montage. F8 group had a positive phase reversal at F3 (open square). Center, the voltage topography of the F8 group showed a tangential dipole pattern (top), whereas that of F4 group showed a radial dipole pattern (bottom). At right, dipole localizations on the coregistered magnetic resonance (MR) image. At top, MR images show the dipoles of the F8 group, and at bottom, those of the F4 group. The images were chosen to demonstrate many dipoles in one slice. same position or close to each other had different orientations when the dipoles were derived from one spike. Our study investigated dipoles in close proximity but with different orientations emanating from two independent spikes. Furthermore, we did not observe any propagation between the tangential and radial dipoles except three of 17 Pz spikes in patient 1. In other words, our results suggest that a large number of neurons in an epileptogenic zone include different groups of neuronal synchronicities that are located in the neighboring gyri but generate interictal spikes independently (Fig. 6A). Baumgartner et al. (13) reported that tangential and radial dipoles represented propagation of epileptiform activity between two adjacent cortical areas (Fig. 6B). Merlet et al. (15), using simultaneously recorded magnetoencephalography (MEG) and EEG source localizations, demonstrated this phenomenon (Fig. 6 B) and showed, further, that an extensive area including fissural and gyral cortices synchronizes immediately when MEG and EEG peaks are simultaneous (Fig. 6C). An abnormal gyrus is difficult to pinpoint with equivalent current dipole analysis because the analysis is a model of complex neuronal activities, uses a relatively small number of electrodes for pediatric monitoring, and has some error of coregistration of EEG dipoles and MR images. However, the coregistered MR images are helpful in understanding an anatomic profile of an epileptic region in part of which tangential and radial dipoles coexist. In our study, the tangential dipoles tended to cluster more than the radial dipoles, as previously reported (6 8,16). The tangential dipole in typical benign rolandic epilepsy of childhood showed a high degree of stability (7). Ebersole and Wade (16) reported that patients with voltage topography of the tangential dipole type had mesial temporal seizures and a better surgical outcome than those with single negative field topography. When voltage topography shows a tangential dipole, equivalent current dipole analysis may lead to stable source localization of interictal spikes, which may indicate the region for surgery. The objective of presurgical evaluation for epilepsy surgery is to investigate the extent of an epileptogenic zone, which is not a pinpoint source but an area of some breadth (17). Equivalent current dipole analysis of individual spikes may contribute to the search. Averaged spikes may hide the actual propagation of epileptic ac-

7 DIPOLE IN INDEPENDENT NEIGHBORING SPIKES 489 FIG. 5. At top, a schema shows the dipole position, orientation, and distance (millimeters) between the dipoles (F8 group, gray circle and black tail; F4 group, black circle and tail) and electrodes in patient 3. Lower schema shows potential distributions along a line on the scalp that were created by actual amplitudes at electrodes of spikes. The amplitude distribution of F8 group (dotted line) shows an S-shaped curve, characteristic of a tangential dipole, consisting of a negative maximum at F8 and a positive maximum at F3. The amplitude distribution of F4 group shows bell-shaped curve (solid line) with only a negative maximum at F4. tivity and, furthermore, make it difficult to delineate the margin of the epileptic area (11,18). The dipole analysis of individual spikes is affected by background EEG activities and artifacts. To solve this problem, one should choose spikes with high signal-to-noise ratio and analyze as many as possible to obtain enough reliable dipoles with high GoF (11,19). The dipole positions of individual interictal spikes in our patients corresponded to the area of cortical excision or multiple subpial transection (MST) based on the results of electrocorticography. Bautista et al. (20) reported that extension of interictal epileptiform discharges beyond the area of resection correlated with poor surgical outcome and that focal discharges inside the surgical resection area correlated with a good outcome in patients with extrahippocampal epilepsy. In patient 3, who had diffuse radial dipoles in the right frontal region, right frontal cortical excision and MST on the premotor area only partially relieved her seizures. In conclusion, when we observe independent neighboring spikes on a patient s EEG, dipole analysis is helpful in understanding and visualizing the neuronal generators of these spikes. A cluster of tangential and radial dipoles in the same epileptic region predicts a good surgical outcome. However, diffuse radial dipoles in the extensive epileptic zone might indicate a poor surgical outcome in patients with extratemporal lobe epilepsy. Acknowledgment: This study was supported by a research grant from the National Epifellows Foundation. We thank Ms. Patti Quint of the Scripps Clinic (San Diego, CA, U.S.A.) for preparing the MRI files in patients 1 and 2 and Dr. William Logan of The Hospital for Sick Children, Toronto, for preparing the MRI files of patient 3. This article was prepared with the assistance of Editorial Services, The Hospital for Sick Children, Toronto. REFERENCES FIG. 6. Three possible mechanisms by which both tangential and radial dipoles exist in an epileptic region. A: Different groups of neuronal generators (open and closed circles) that discharge independently are located in neighboring gyri. B: Propagation: early epileptic activity arises from a fissural cortex, and then spreads to a crown of a gyrus. C: An extensive area including several fissural and gyral cortices synchronizes immediately. 1. Gloor P. Neuronal generators and the problem of localization in electroencephalography: application of volume conductor theory to electroencephalography. J Clin Neurophysiol 1985;2: Bishop GH. Potential phenomena in thalamus and cortex. Electroenceph clin Neurophysiol 1949;1: Lesser RP, Lüders H, Dinner DS, et al. An introduction to the basic concepts of polarity and localization. J Clin Neurophysiol 1985;2: Burgess RC, Collura TF. Polarity, localization, and field determination in electroencephalography. In: Wyllie E, ed. The treatment of epilepsy: principles and practice. Philadelphia: Lea & Febiger, 1993: Lopes da Silva FH. A critical review of clinical applications of topographic mapping of brain potentials. J Clin Neurophysiol 1990;7: Gregory DL, Wong PKH. Clinical relevance of a dipole field in rolandic spikes. Epilepsia 1992;33: Wong PKH. Stability of source estimates in rolandic spikes. Brain Topogr 1989;2: Wong PKH. Source modeling of the rolandic focus. Brain Topogr 1991;4: Van der Meij W, Wieneke GH, Van Huffelen AC. Dipole source analysis of rolandic spikes in benign rolandic epilepsy and other clinical syndromes. Brain Topogr 1993;5:

8 490 A. OCHI ET AL. 10. Yamazaki T, Kamijo K, Kenmochi A, et al. Multiple equivalent current dipole source localization of visual event-related potentials during oddball paradigm with motor response. Brain Topogr 2000; 12: Ochi A, Otsubo H, Shirasawa A, et al. Systematic approach to dipole localization of interictal EEG spikes in children with extratemporal lobe epilepsies. Clin Neurophysiol 2000;111: Ebersole JS. Defining epileptogenic foci: past, present, future. J Clin Neurophysiol 1997;14: Baumgartner C, Graf M, Doppelbauer A, et al. The functional organization of the interictal spike complex in benign rolandic epilepsy. Epilepsia 1996;37: Boon P, D Havé M, Adam C, et al. Dipole modeling in epilepsy surgery candidates. Epilepsia 1997;38: Merlet I, Paetau R, García-Larrea L, Uutela K, et al. Apparent asynchrony between interictal electric and magnetic spikes. Neuroreport 1997;8: Ebersole JS, Wade PB. Spike voltage topography identifies two types of frontotemporal epileptic foci. Neurology 1991;41: Lüders HO, Awad I. Conceptual considerations. In: Lüders HO, ed. Epilepsy surgery. New York: Raven Press, 1991: Ko DY, Kufta C, Scaffidi D, et al. Source localization determined by magnetoencephalography and electroencephalography in temporal lobe epilepsy: comparison with electrocorticography: technical case report. Neurosurgery 1998;42: Watanabe Y, Sato S, Nakamura F, et al. The practical benefits of magnetoencephalography in comparison with electroencephalography in a patient with epilepsia partialis continua. No To Sinkei 1995;47: [Japanese] 20. Bautista RE, Cobbs MA, Spencer DD, et al. Prediction of surgical outcome by interictal epileptiform abnormalities during intracranial EEG monitoring in patients with extrahippocampal seizures. Epilepsia 1999;40: