High-resolution EEG: Cortical potential imaging of interictal spikes

Transcription

1 Clinical Neurophysiology 114 (2003) High-resolution EEG: Cortical potential imaging of interictal spikes X. Zhang a, W. van Drongelen b,c, K.E. Hecox b, V.L. Towle c,d, D.M. Frim d, A.B. McGee b,b.he a, * a Department of Bioengineering, The University of Illinois at Chicago, MC-063, SEO 218, 851 South Morgan Street, Chicago, IL, USA b Department of Pediatrics, The University of Chicago, Chicago, IL, USA c Department of Neurology, The University of Chicago, Chicago, IL, USA d Department of Surgery, The University of Chicago, Chicago, IL, USA See Editorial, pages Abstract Background: It is of clinical importance to localize pathologic brain tissue in epilepsy. Noninvasive localization of cortical areas associated with interictal epileptiform spikes may provide important information to facilitate presurgical planning for intractable epilepsy patients. Methods: A cortical potential imaging (CPI) technique was used to deconvolve the smeared scalp potentials into the cortical potentials. A 3-spheres inhomogeneous head model was used to approximately represent the head volume conductor. Five pediatric epilepsy patients were studied. The estimated cortical potential distributions of interictal spikes were compared with the subsequent surgical resections of these same patients. Results: The areas of negativity in the reconstructed cortical potentials of interictal spikes in 5 patients were consistent with the areas of surgical resections for these patients. Conclusions: The CPI technique may become a useful alternative for noninvasive mapping of cortical regions displaying epileptiform activity from scalp electroencephalogram recordings. q 2003 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved. Keywords: Cortical imaging; High-resolution EEG; Epilepsy; Source localization; Presurgical planning; Inverse problem 1. Introduction The electroencephalogram (EEG) has unsurpassed millisecond temporal resolution and it is an economic and easy-to-use diagnostic modality. However, the poor spatial resolution of the conventional EEG limits its use for mapping and imaging spatially distributed brain electrical activity. In recent years, tremendous effort has been made to enhance the spatial resolution of the conventional EEG by means of high-resolution EEG approaches, which largely overcome the head volume conductor effect (Cohen and Cuffin, 1990; Sidman et al., 1990; Dale and Sereno, 1993; Ebersole, 1994; Gevins et al., 1994; Nunez et al., 1994; He et al., 1996, 1999, 2001b, 2002; Babiloni * Tel.: þ ; fax: þ address: bhe@uic.edu (B. He). et al., 1997, 2000; Fuchs et al., 1999; He, 1999; Michel et al., 1999; Scherg et al., 1999). Of particular interest is the recent development of the cortical potential imaging (CPI) technique, in which an explicit biophysical model of the passive conducting components of the head is used to deconvolve a recorded scalp potential distribution into a distribution of electrical potential over the epicortical surface (for review, see He, 1999). There are two approaches in CPI, the indirect CPI and the direct CPI. In the indirect CPI approach (Sidman et al., 1990; He et al., 1996, 2001b; Babiloni et al., 1997; Wang and He, 1998; He, 1998, 1999), the cortical potentials are constructed using an intermediate equivalent dipole layer. In such an approach, the inverse procedure is used to estimate the equivalent dipole distribution from the scalp EEG, and then the cortical potentials are reconstructed by solving the forward problem, from the estimated equivalent dipole /03/$30.00 q 2003 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved. doi: /s (03)

2 1964 X. Zhang et al. / Clinical Neurophysiology 114 (2003) layer to the cortical potentials. In the direct CPI approach, investigators directly linked the scalp potentials with the cortical potentials by means of the finite element method (FEM), the boundary element method (BEM), or the spherical harmonics (Le and Gevins, 1993; Srebro et al., 1993; Gevins et al., 1994; Nunez et al., 1994; Edlinger et al., 1998; He et al., 1999, 2002; Ollikainen et al., 2001). Both simulation and human experimental studies have demonstrated that more localized spatial distributions of cortical potentials can be obtained compared to the spatial distribution of scalp potentials (He et al., 2002). In the clinical treatment of epilepsy, it is of importance to locate intracranial generators of interictal and ictal epileptiform activity. The electrocorticogram (ECoG) recorded from subdural grid electrodes has become the gold standard for defining epileptogenic brain regions (Engel et al., 1981). However, invasiveness, risk of morbidity, limited availability, and high cost of this technique limit its routine clinical use (Huppertz et al., 2001a). Many methods have been used for presurgical evaluation of epilepsy patients (Ebersole, 1994; Diekmann et al., 1998; Huppertz et al., 2001a,b). The dipole localization method, as well as the distributed source analysis method, has become commonly used in both research and clinical studies for predicting the intracranial sources. On the other hand, CPI has not been widely used in this situation, with only two case reports describing the application of CPI for mapping the cortical regions displaying epileptiform activity (Sidman et al., 1992; Gevins et al., 1999). The objective of the present study was to test the feasibility of localizing and imaging cortical regions displaying interictal epileptiform activity from clinical pre-operative scalp EEG recordings in pediatric epilepsy patients by means of CPI. The estimated CPI results were validated by comparing them with the subsequent surgical resections in the same patients and the outcomes of the surgical resections. 2. Method 2.1. Patients Five pediatric patients (one male, 4 females, 7 16 years old, Table 1) with intractable epilepsy were studied using a protocol approved by the Institutional Review Boards of the University of Illinois at Chicago and The University of Chicago. The clinical studies took place at the Pediatric Epilepsy Center at The University of Chicago Children s Hospital. All patients had similar evaluations that included structural MRI, interictal, and ictal long-term video EEG monitoring, with scalp electrodes and subsequently with subdural electrodes (Radionics Medical Products Inc., Burlington, Massachusetts). After surgical resection, all patients are either seizure free or have had substantial reduction of seizure frequency (Table 1) Data acquisition All scalp EEG recordings were obtained during presurgical monitoring. The scalp EEG data were obtained from 24 scalp electrodes, placed according to the system (American Electroencephalographic Society, 1994) with a sampling rate of 400 Hz and band-pass filtering at Hz (BMSI 6000, Nicolet Biomedical Inc., Wisconsin). The interictal records from each patient were reviewed for occurrence of spikes. Interictal spikes were identified via visual inspection according to IFSECN criteria (Chatrian et al., 1974). Spikes were selected sequentially as they occurred in the record. Epochs with artifact (eye movement and blinks, muscle activity, etc.) were excluded from analysis. Using these criteria, at least 6 interictal spikes were selected for each patient. The artifact-free pre-operative scalp EEG epochs of 2 s duration, including the interictal spikes, were selected and the highest negative peak of the interictal spike was selected by a peak detection algorithm. Baseline-correction was based on the scalp EEG data from Table 1 Diagnosis, results of presurgical evaluations, and post-operative outcome for all 5 patients Patient Age Gender Diagnosis Operation No. of resections Outcome 1 13 Female Intractable epilepsy with left temporal focus 2 8 Female Intractable epilepsy with left temporal focus 3 7 Female Intractable epilepsy with a right medial temporal focus and a right posterior temporal parietal focus 4 16 Male Intractable epilepsy with left front focus 5 8 Female Intractable epilepsy with left frontal lobe seizure focus Left temporal lobectomy 1 Seizure free Left temporal lobectomy and left parietal topectomy Right temporal lobectomy and right parietal topectomy 3 Seizure free 2 Seizure free Frontal lobectomy 1 Reduction of seizure occurrence Left frontal lobectomy 1 Seizure free

3 X. Zhang et al. / Clinical Neurophysiology 114 (2003) to 100 ms before the negative peak of the interictal discharge. In order to capture both the early onset and the peak of the epileptiform activity, CPI was applied to time points from 20 ms before to 5 ms after the negative peak Cortical potential imaging The CPI technique used in the present study has been previously tested in computer simulation and validated in an SEP experimental study in humans (He et al., 1999, 2002). In brief, the head was approximately represented by a 3-shell volume conductor with the 3 shells representing the scalp, the skull, and the brain, respectively. Each shell is homogeneous but has different electrical conductivity. Since brain electrical sources exist only inside the brain, Green s second identity can be applied to the volume between the scalp and the skull and the volume between the skull and the brain, separately. After mathematical manipulations, the cortical potential U 3 over the epicortical surface can be directly related to the scalp potential U 1 by: U 1 ¼ T 13 U 3 ð1þ where T 13 is the transfer matrix from the cortical potential to the scalp potentials. In practice, the vector of the measured scalp potentials F is a subset of the potential vector U 1. Therefore, F can be connected with the cortical potentials by: F ¼ AU 3 ð2þ where A is the sub-matrix of T 13. To account for the lowconductivity skull layer, an adaptive approach has been developed to achieve high numerical accuracy in the transfer matrix A (He et al., 1999). The inverse problem of CPI is to seek the unknown cortical potential U 3 from the measured scalp potential F. To overcome the ill-posed nature of the inverse problem, zero-order Tikhonov regularization was applied (Tikhonov and Arsenin, 1977): U 3 ¼ðA T A þ liþ 21 A T F where l is the regularization parameter, which is used to suppress the effect of noise. The determination of the regularization parameter was achieved by means of the discrepancy principle (Morozov, 1984), in which the regularization parameter is determined such that the following equation is satisfied, kau 3 2 Fk ¼ knk, where knk is the noise level in the measurements. In the present study, the head volume conductor was approximately represented using an inhomogeneous 3-concentric-sphere model, in which the radii of the brain, the skull, and the scalp spheres were taken as 0.87, 0.92, and 1.0, respectively. The normalized conductivity of the scalp and the brain was taken as 1.0, and that of the skull as 1/15 (Oostendorp et al., 2000). For the present study, a spherical model and standard electrode locations were used in order to test the feasibility of using the CPI approach in clinical ð3þ routines where a whole-head MRI and measured electrode positions are not available (as is frequently the case in a clinical setting). The coordinates of the standard electrode locations were determined by projecting the electrode positions of the system onto a spherical surface. After obtaining the CPI results in a spherical model, the results were also illustrated over a standard realistic geometry (RG) head model by projecting the spherical CPI results onto the standard RG cortical surface. This projection was accomplished by 3 steps. First, the spherical model was set to the physical scale of the RG model, the optimal radius and center of sphere were obtained by minimizing the norm of error distance between the boundary elements of the RG model and the spherical model. Second, the spherical CPI results were interpolated onto the RG cortical surface model using a RG spline interpolation algorithm (He et al., 2001a). Then, the elements on the RG cortical surface model which exceeded 50% of the maximum negative value of the estimated cortical potential were marked, and one (or multiple) circle(s) or ellipse(s) was (were) drawn to cover all the marked elements to illustrate the area(s) of epileptiform activity. 3. Results 3.1. Patient 1 Patient 1 is a 13-year-old female with intractable epilepsy. The pre-operative long-term scalp EEG monitoring showed that this patient had a left temporal epileptic focus. Pre-operative long-term subdural EEG monitoring revealed that this patient had seizure foci on her lateral and medial temporal lobe structures of the left hemisphere. The patient was seizure free after left temporal lobectomy. Six interictal spikes were chosen and CPI was performed on each spike. Fig. 1 shows a typical example of the CPI results for this patient. Figs. 1a and b show left-side scalp potential distributions and cortical potential distributions estimated at different time points, during the interictal spike marked by the red dot in Fig. 1c. The 0 ms represents the time point of the highest negative peak on T7, marked by the red dot in Fig. 1c. The smeared and distorted scalp potential distributions suggest that the epileptiform activity might have originated in the left temporal lobe and propagated to the whole left hemisphere. However, in Fig. 1b, with enhanced spatial resolution, the CPI results suggest that there were two regions of negative activity: N1 over the left anterior temporal region and N2 at the left posterior temporal region. Almost the whole temporal region then became involved, at the peak of the interictal spike. Fig. 1d shows the projection of the CPI result at 210 ms onto a standard cortical surface model, which illustrates the two areas of negativity in the estimated CPI images. The N1 and N2 activities appeared in all CPI analyses of 6 interictal spikes from this patient.

4 1966 X. Zhang et al. / Clinical Neurophysiology 114 (2003) Fig. 1. Patient 1. (a) Scalp potential distributions at different time points. (b) Estimated cortical potential distributions at correspondent time points. The black dot represents the position of nose. (c) Interictal spike waveform. The red dot represents the highest negative peak. (d) Projection of cortical potential imaging results onto a standard cortical surface model for visual interpretation. The lobectomy extended approximately 4 cm from the temporal pole and then turned down toward the temporal floor, all the cortex and white matter along the inferior bank of the sylvian fissure all the way to the temporal tip was removed as well as all the far frontal superior temporal gyrus tissue. The N1 activity in our CPI results appears to overlap well with the resection. This patient also underwent dissection of the temporal area (including the hippocampus) all the way back to behind the curvature of the brain stem. The N2 activity in our CPI results appears to overlap with a part of the second resection area Patient 2 Patient 2 is an 8-year-old female with intractable seizures. The pre-operative long-term scalp EEG monitoring revealed interictal discharges over the left hemisphere. Four subdural electrode grids were implanted to cover the posterior part of the left frontal lobe, the left parietal lobe, the left posterior parietal temporal lobe, and the left superior and anterior temporal lobe. Pre-operative longterm subdural EEG monitoring revealed 3 areas with seizure discharges, one over the left parietal lobe and two over the left temporal lobe. Neurosurgical resection of these areas (left parietal topectomy and left temporal lobectomy) made the patient seizure free. Nine pre-operative interictal spikes were chosen for the CPI analysis. Figs. 2a and b show two typical examples of scalp potential maps and Figs. 2c and d show the corresponding CPI results obtained from two interictal spikes. Substantial enhancement in spatial details was obtained by means of CPI. Three localized areas of negativity, one over the left parietal region (N1) and two over the left temporal region (N2 or N3) were observed. Among all 9 interictal spikes, the CPI results show N1 activity in all cases, while N2 and N3 activities appear in only 5 and 3 cases, respectively. Three resections were performed on this patient, as illustrated in Fig. 2e. The first resection was conducted at an area across the sylvian fissure from the inferior frontal lobe to the superior temporal lobe. The N3 activity appears to overlap with this resection area. Note that the N1 area is associated with an area of positivity slightly to anterior and superior. This can be an indication for a tangentially oriented dipole located superficially in between the two extrema. The second topectomy was on the parietal lobe, including the area of the N1 activity. Another subpial transection was performed over the posterior temporal lobe, which was overlapped by N2 activity. The locations of the 3 resections (Fig. 2e) are consistent with the 3 negative activities estimated from the CPI analysis (Fig. 2f) Patient 3 Patient 3 is a 7-year-old female with intractable seizures. Pre-operative long-term scalp EEG monitoring implicated

5 X. Zhang et al. / Clinical Neurophysiology 114 (2003) Fig. 2. Patient 2. (a and b) Spike waveforms of two selected interictal spikes and associated right-sided and top views of scalp potential maps. The black dot represents the position of nose. The red dot represents the highest negative peak of interictal spike. (c and d) Right-sided and top views of cortical potential imaging results corresponding to (a and b). (e) Lateral view of the surgical resections. (f) Projection of cortical potential imaging results onto a standard cortical surface model for visual interpretation.

6 1968 X. Zhang et al. / Clinical Neurophysiology 114 (2003) Fig. 3. Patient 3. (a) Right-sided and top views of scalp potential maps. (b) Right-sided and top views of estimated cortical potential distributions at corresponding time points. The black dot represents the position of nose. (c) Interictal spike waveform. The red dot represents the highest negative peak. (d) Lateral view of the surgical resections. (e) Projection of cortical potential imaging results onto a standard cortical surface model for visual interpretation. (f) Temporal behavior of the strengths of N1 and N2 activities. right side seizure activity. Subdural electrode grids were implanted over the right temporal lobe, the right parietal lobe, and the right posterior part of the temporal lobe. The pre-operative long-term subdural EEG monitoring revealed two seizure foci involving the right medial temporal lobe and the right parietal lobe. The neurosurgical resection (right temporal lobectomy and right parietal topectomy) made the patient seizure free.

7 X. Zhang et al. / Clinical Neurophysiology 114 (2003) Seven pre-operative interictal spikes were chosen and subjected to CPI analysis. Fig. 3 shows a typical example. From the waveforms of the interictal spike (Fig. 3c), the highest negative peak appeared at P4 and another smaller peak appeared at P8. Figs. 3a and b show the scalp potential maps and estimated cortical potential maps (right view and top view). Compared with smeared scalp potential distributions, the estimated cortical potential distributions clearly show that the N1 activity originated at the right parietal region (217.5 ms) and a second area of activity (N2) originated at the right temporal region shortly after the N1 activity appeared. Fig. 3f depicts the amplitudes of N1 and N2 activities versus time. Both activities persisted for the duration of the spike onset. The N2 activity diminished before the N1 activity. Both the N1 and N2 activities were present in all 7 interictal spikes. Fig. 3e illustrates the projection of the N1 and N2 areas onto a standard cortical surface model, showing two separate activities in the right parietal region and the right temporal region. The CPI results are consistent with the surgical resection, illustrated in Fig. 3d. The right parietal topectomy was performed from the anterior posterior level of the vein of Labbe extending posteriorly for two gyri. This resection area was consistent with the N1 activity. The right temporal lobectomy was limited superiorly by the sulcus between the superior temporal gyrus and the middle temporal gyrus, and posteriorly by the vein of Labbe. The resected area corresponded with the N2 activity Patient 4 Patient 4 is a 16-year-old male with intractable epilepsy. The pre-operative long-term scalp EEG monitoring suggested presence of epilepsy foci in the frontal part of the brain. Subdural electrode grids were implanted to cover the left frontal lobe and frontally in between the hemispheres. The pre-operative long-term subdural EEG monitoring showed a seizure focus on the left frontal lobe. After a left frontal lobectomy, seizures in this patient were reduced from several per day to one seizure every 1 2 weeks. Thirteen pre-operative interictal spikes were chosen and subjected to CPI analysis and the typical examples are shown in Fig. 4. Fig. 4a and b, the left and top views of CPI results, show that the epileptiform activity originated from the left superior frontal region (212.5 ms) and propagated toward the medial face (25 to 5 ms), as well as a posterior from there (0 5 ms). The surgical resection included the inferior frontal gyrus and the entire lateral and medial surfaces of the frontal lobe. The resection extended to the posterior and superior margin of a previous resection, where some evidence of spiking was found on subdural interhemispheric electrodes. Figs. 4c and d depict the waveforms of the interictal spike and the projection of the areas of negativity onto a standard cortical surface model, respectively. The areas of negativity in the CPI results appear to be consistent with the surgical resection Patient 5 Patient 5 is an 8-year-old female with intractable seizures. Pre-operative long-term scalp EEG suggested a left frontal lobe origin. Electrode grids were implanted over the left frontal lobe, the left temporal lobe, and the interhemisphere left and right frontal lobes for subdural recording. The pre-operative long-term subdural EEG monitoring revealed that a large portion of left frontal lobe was involved in seizure activity. A left frontal lobectomy made the patient seizure free. Ten pre-operative interictal spikes were chosen and subjected to CPI analysis. From the waveforms of interictal spike (Fig. 5c), it is seen that the highest negative peak was on F3, with FP1 and F7 also having strong activity, as well as small spikes at Fz and T7. Figs. 5a and b show that the epileptiform activity was spread widely on the left frontal region and was consistent with the surgical resection (Fig. 5d). 4. Discussion In the present study, 5 pediatric patients pre-operative interictal spikes were subjected to cortical potential image analysis using a 3-concentric sphere head model. The estimated cortical potentials revealed localized areas of negativity over the epicortical surface, which were consistent with the resected cortical areas of the patients. The present study extends the investigations of others (Assaf and Ebersole, 1997; Huppertz et al., 2001b) to extratemporal epilepsy. For pediatric patients, the epilepsy frequently involves different areas of the neocortex and a greater variety in patterns of EEG and clinical findings as compared with adult patients (Gilliam et al., 1997; Otsubo et al., 1997; Ochi et al., 2000). In the present study, we have been able to successfully reveal epileptiform activity associated with different cortical areas including the frontal, parietal, and temporal lobes. CPI is a surface imaging technique which images the electrical potential distribution over the epicortical surface, that would be recorded if electrodes were placed over the same surface. We applied CPI to evaluate cortical topography of brain activity associated with interictal spikes. Although this method does not calculate the depth of the source, it may identify regions that generate or participate in epileptiform activity. Our interest to explore electricity at the cortical surface relates to our pediatric patient group, where the epileptic focus (or foci) is often found in different parts of the neocortex including extratemporal areas (unlike in adults). The present promising results suggest that such inversely estimated cortical potential distribution may be useful in determining cortical regions associated with epileptiform activity, due to the physical vicinity of the cortical surface to underlying cortical sources. Comparison between the scalp potential distributions and the estimated CPI

8 1970 X. Zhang et al. / Clinical Neurophysiology 114 (2003) Fig. 4. Patient 4. (a) Left-sided view of the estimated cortical potential distributions at different time points. (b) Top view of estimated cortical potential distributions at corresponding time points. The black dot represents the position of nose. (c) Interictal spike waveform. The red dot represents the highest negative peak. (d) Projection of cortical potential imaging results onto a standard cortical surface model for visual interpretation. results clearly indicates the enhanced spatial details in the CPI results. This spatial enhancement provides useful information for assessing and imaging the underlying cortical activity, despite the fact that the CPI result is an electrical potential itself. The CPI technique is limited in that it lacks the depth information on the sources generating epileptiform activity. The advantage of our approach as compared to the dipole methods is that an a priori estimate of the number of active regions is not required. This is a key feature of CPI since an important question in presurgical evaluation of patients with intractable epilepsy is how many cortical areas are involved in epileptiform activity. In addition, the CPI approach has the benefit that the results can easily be interpreted by neurologists in the clinical decision process since it is estimation of the ECoG that would be recorded should an electrode array be placed on the cortical surface. Also, in the pediatric patient group where the epileptic activity is often located in neocortex, the noninvasive capability of estimating ECoG over the epicortical surface would have potential of guiding surgical planning. In clinical recordings with subdural electrodes, the localizations are frequently occurring closer to the edge of the recording array as one would like. Since the CPI technique uses all information recorded from all the electrodes, such an edge effect does not seem to exist. Rather, it appears to be a unique feature of the present CPI approach, as it is desirable in a clinical setting to be able to predict various cortical regions displaying epileptiform activity from limited number of scalp electrodes. Individual interictal spikes were selected for CPI in the present study. Although spike averaging can increase

9 X. Zhang et al. / Clinical Neurophysiology 114 (2003) Fig. 5. Patient 5. (a) Left-sided view of cortical potential imaging results at different time points. (b) Top view of estimated cortical potential distributions at corresponding time points. The black dot represents the position of nose. (c) Interictal spike waveform. The red dot represents the highest negative peak. (d) Projection of cortical potential imaging results onto a standard cortical surface model for visual interpretation. the signal-to-noise ratio, it also may lead to estimation error (Baumgartner et al., 1995; Lantz et al., 1997; Diekmann et al., 1998; Michel et al., 1999). It is well known that pediatric patients exhibit variability of the waveform of individual spikes, implying that single spike analysis is preferable whenever possible (Michel et al., 1999; Ochi et al., 2000). As an example, for patient 2 (Fig. 2), it is shown that by performing single spike CPI, the N2 activity was revealed in 3 of 9 interictal spikes and N3 was revealed for 5 of 9 interictal spikes. Such dynamic behaviors might be overlooked if the analysis was performed on the averaged data. In patients 1, 3, 4, and 5, the cortical regions identified for each patient were consistent across different interictal spikes. The results presented in Figs. 1 5 represent typical examples among the spikes analyzed. Also shown in Fig. 3f is the dynamic change in amplitude of the negativity N1 and N2 in patient 3. It is of importance for presurgical planning to know the exact location of the epileptic region, as well as the size and shape of epileptogenic zones. However, there is currently no single widely accepted method to achieve this goal. It has been shown that for dipole localization, the average error between the calculated dipole and the intracranial source was 8.5 mm for a single dipole model using a 3 compartment RG head model (Homma et al., 1994) and 10 mm for a 4 compartment spherical model (Cohen et al., 1990). Another study showed that the average difference of source localizations with the 3-sphere and RG head model was about 2 cm (Roth et al., 1997). A previous study (Silva et al., 1999) has reported that use of RG head models increased the dipole localization accuracy for epileptiform spikes as compared to the use of spherical head model. For distributed source modeling, there is no certain criterion to quantitatively evaluate the localization results. A recent study (Lantz et al., 2003) suggested that sufficient spatial sampling is important to obtain the accurate source localization when a distributed source imaging approach is used to analyze the epileptiform activity. In the present study, we attempted to image and localize the cortical regions displaying epileptiform activity based on the pre-operative scalp EEG recordings made in a clinical setting. For this purpose, we used the 3-concentric-spheres inhomogeneous head modeling and standardized electrode positions to test the feasibility of applying CPI to clinical data where no complete structural MRI and measured electrode positions are available. We did not attempt to obtain exact locations of the cortical regions covering epileptogenic zones, but rather an indication of the areas of the cortex under which epileptogenic zones may exist. For this purpose, the CPI results obtained in the 3-spheres head model have been projected onto a standard cortical surface, in order to provide a qualitative comparison with the neurosurgical

10 1972 X. Zhang et al. / Clinical Neurophysiology 114 (2003) resections. For accurate localization purpose, RG head models should be used in the inverse computation. Nevertheless, the promising results of the present study suggest the feasibility of imaging and localizing cortical regions displaying epileptiform activity. The results suggest that one can noninvasively identify the cortical regions that generate the interictal epileptiform activity, the propagation of that activity, and the extent of the irritative zone. Such information would be useful to aid clinical decision making for presurgical planning of subdural grid placement for epilepsy patients. Although beyond the scope of this work, future investigations should address accurate localization by using RG head models built from individual patients MR images and evaluation of the effects of accurate head modeling and spatial sampling on the CPI results. In summary, we have tested the feasibility of imaging cortical regions displaying epileptiform activity by means of CPI analysis of interictal spikes from 5 pediatric epilepsy patients. Without a priori knowledge, the estimated cortical potential maps successfully revealed localized areas of activity overlapping with epileptogenic zones as identified from clinical findings and confirmed by surgical resections. The present study suggests that CPI may become an alternative means for noninvasive localization of cortical regions displaying epileptiform activity and may potentially be useful for presurgical planning in pediatric epilepsy patients. References American Electroencephalographic Society. American Electroencephalographic Society. Guideline thirteen: guidelines for standard electrode position nomenclature. J Clin Neurophysiol 1994;11: Assaf BA, Ebersole JS. Continuous source imaging of scalp ictal rhythms in temporal lobe epilepsy. Epilepsia 1997;38: Babiloni F, Babiloni C, Carducci F, Fattorini L, Anello C, Onorati P, et al. High resolution EEG: a new model-dependent spatial deblurring method using a realistically-shaped MR-constructed subject s head model. Electroenceph clin Neurophysiol 1997;102: Babiloni F, Carducci F, Cincotti F, Del Gratta C, Roberti GM, Romani GL, et al. Integration of high resolution EEG and functional magnetic resonance in the study of human movement-related potentials. Methods Inf Med 2000;39: Baumgartner C, Lidinger G, Ebner A, Aull S, Serles W, Olbrich A, et al. Propagation of interictal epileptic activity in temporal lobe epilepsy. Neurology 1995;45: Chatrian GE, Bergamini L, Dondey M, Klass DW, Lennox-Buchtal M, Petersen I. A glossary of term most commonly used by clinical electroencephalographers. Electroenceph clin Neurophysiol 1974;37: Cohen D, Cuffin BN, Yunokuchi K, Maniewski R, Purcell C, Cosgrove GR, et al. MEG versus EEG localization test using implanted sources in the human brain. Ann Neurol 1990;28: Dale AM, Sereno MI. Improved localization of cortical activity by combining EEG and MEG with MRI cortical surface reconstruction: a linear approach. J Cogn Neurosci 1993;5: Diekmann V, Becker W, Jürgens R, Grozinger B, Kleiser B, Richter HP, et al. Localisation of epileptic foci with electric, magnetic and combined electromagnetic models. Electroenceph clin Neurophysiol 1998;106: Ebersole JS. Non-invasive localization of the epileptogenic focus by EEG dipole modeling. Acta Neurol Scand 1994;152:20 8. Edlinger G, Wach P, Pfurtscheller G. On the realization of an analytic highresolution EEG. IEEE Trans Biomed Eng 1998;45: Engel Jr J, Rausch R, Lieb JP, Kuhl DE, Crandall PH. Correlation of criteria used for localizing epileptic foci in patient considered for surgical therapy of epilepsy. Ann Neurol 1981;9: Fuchs M, Wagner M, Kohler T, Wischmann HA. Linear and nonlinear current density reconstructions. J Clin Neurophysiol 1999;16: Gevins A, Le J, Martin NK, Brickett P, Desmond J, Reutter B. High resolution EEG: 124-channel recording, spatial deblurring and MRI integration method. Electroenceph clin Neurophysiol 1994;90: Gevins A, Le J, Leong H, McEvoy L, Smith M. Deblurring. J Clin Neurophysiol 1999;16: Gilliam F, Wyllie E, Kashden J, Fraught E, Kotagal P, Bebin M, et al. Epilepsy surgery outcome: comprehensive assessment in children. Neurology 1997;48: He B. High resolution source imaging of brain electrical activity. IEEE Eng Med Biol Mag 1998;17: He B. Brain electric source imaging: scalp Laplacian mapping and cortical imaging. Crit Rev Biomed Eng 1999;27: He B, Wang Y, Pak S, Ling Y. Cortical source imaging from scalp electroencephalograms. Med Biol Eng Comput 1996;34: He B, Wang Y, Wu D. Estimating cortical potentials from scalp EEG s in a realistically shaped inhomogeneous head model. IEEE Trans Biomed Eng 1999;46: He B, Lian J, Li G. High-resolution EEG: a new realistic geometry spline Laplacian estimation technique. Clin Neurophysiol 2001a;112: He B, Lian J, Spencer KM, Dien J, Donchin E. A cortical potential imaging analysis of the P300 and novelty P3 components. Hum Brain Mapp 2001b;12: He B, Zhang X, Lian J, Sasaki H, Wu D, Towle VL. Boundary element method-based cortical potential imaging of somatosensory evoked potentials using subjects magnetic resonance images. Neuroimage 2002;16: Homma S, Musha T, Nakajima Y, Okamoto Y, Blom S, Flink R, et al. Location of electric current sources in the human brain estimated by the dipole tracing method of the scalp skull brain (SSB) head model. Electroenceph clin Neurophysiol 1994;91: Huppertz HJ, Hof E, Klisch J, Wagner M, Lücking CH, Kristeva-Feige R. Localization of interictal delta and epileptiform EEG activity associated with focal epileptogenic brain lesions. Neuroimage 2001a;13: Huppertz HJ, Hoegg S, Sick C, Lücking CH, Zentner J, Schulze-Bonhage A, et al. Cortical current density reconstruction of interictal epileptiform activity in temporal lobe epilepsy. J Clin Neurophysiol 2001b;112: Lantz G, Michel CM, Pascual-Marqui RD, Spinelli L, Seeck M, Seri S, et al. Extracranial localization of intracranial interictal epileptiform activity using LORETA (low resolution electromagnetic tomography). Electroenceph clin Neurophysiol 1997;102: Lantz G, Grave de Peralta R, Spinelli L, Seeck M, Michel CM. Epileptic source localization with high density EEG: how many electrodes are needed? Clin Neurophysiol 2003;114(1):63 9. Le J, Gevins A. Method to reduce blur distortion from EEG s using a realistic head model. IEEE Trans Biomed Eng 1993;40: Michel CM, de Peralta RG, Lantz G, Andino SG, Spinelli L, Blanke O, et al. Spatiotemporal EEG analysis and distributed source estimation in presurgical epilepsy evaluation. J Clin Neurophysiol 1999;16: Morozov VA. Methods for solving incorrectly posed problems. Berlin: Springer-Verlag; Nunez P, Silibertein RB, Cdush PJ, Wijesinghe RS, Westdrop AF, Srinivasan R. A theoretical and experimental study of high resolution EEG based on surface Laplacian and cortical imaging. Electroenceph clin Neurophysiol 1994;90:40 57.

11 X. Zhang et al. / Clinical Neurophysiology 114 (2003) Ochi A, Otsubo H, Shirasawa A, Hunjan A, Sharma R, Bettings M, et al. Systematic approach to dipole localization of interictal EEG spikes in children with extratemporal lobe epilepsies. J Clin Neurophysiol 2000; 111: Ollikainen JO, Vaukhonen M, Marko V, Karjalainen PA, Kaipio JP. A new computational approach for cortical imaging. IEEE Trans Med Imaging 2001;20: Oostendorp TF, Delbeke J, Stegeman DF. The conductivity of the human skull: results of in vivo and in vitro measurements. IEEE Trans Biomed Eng 2000;47: Otsubo H, Steinlin M, Hwang PA, Sharma R, Jay V, Becker LE, et al. Positive epileptiform discharge in children with neuronal migration disorders. Pediatr Neurol 1997;16: Roth BJ, Ko D, von Albertini-Carletti IR, Scaffidi D, Sato S. Dipole localization in patients with epilepsy using the realistically shaped head model. Electroenceph clin Neurophysiol 1997;102: Scherg M, Bast T, Berg P. Multiple source analysis of interictal spikes: goals, requirements, and clinical value. J Clin Neurophysiol 1999;16: Sidman R, Ford M, Ramsey G, Schlichting C. Age-related features of the resting and P300 auditory evoked responses using the dipole localization method and cortical imaging technique. J Neurosci Methods 1990;33: Sidman R, Vincent DJ, Smith DB, Lee L. Experimental tests of the cortical imaging technique applications to the response to median nerve stimulation and the localization of epileptiform discharge. IEEE Trans Biomed Eng 1992;39: Silva C, Almeida R, Oostendorp T, Ducla-Soares E, Foreid JP, Pimentel T. Interictal spike localization using a standard realistic head model: simulations and analysis of clinical data. Clin Neurophysiol 1999;110: Srebro R, Oguz RM, Hughlett K, Purdy PD. Estimating regional brain activity from evoked potential field on the scalp. IEEE Trans Biomed Eng 1993;40: Tikhonov AN, Arsenin VY. Solution of ill-posed problems. New York, NY: Wiley; Wang Y, He B. A computer simulation study of cortical imaging from scalp potentials. IEEE Trans Biomed Eng 1998;45: