complexidade e epilepsia

Epilepsia, 41(7):81 1-817, 2000 Lippincott Williams & Wilkins, Inc., Baltimore 0 International League Against Epilepsy

Clinical Research

Spatial Distribution of Neuronal Complexity Loss in Neocortical Lesional Epilepsies

G. Widman, *K. Lehnertz, TH. Urbach, and *C. E. Elger

Clinic ,for Neurology, University Hospital Essen; *Department of Epileptology and fDepartment of Neuroradiology, Medical Center, University of Bonn, Germany

Summary: Purpose: Nonlinear EEG analysis is valuable in characterizing the spatiotemporal dynamics of the epileptogenic process in mesial temporal lobe epilepsy. We examined the ability of the measure neuronal complexity loss (L*) to characterize the primary epileptogenic area of neocortical lesional epilepsies during the interictal state.

Methods: Spatial distribution of L* (L* map) was extracted from electrocorticograms (n = 52) recorded during presurgical assessment via subdural 64-contact grid electrodes covering lesions in either frontal, parietal, or temporal neocortex in 15 patients. The exact location of recording contacts on the brain surface was identified by matching a postimplant lateral x-ray of the skull with a postoperatively obtained sagittal MRI scan. Reprojecting L* maps onto the subject’s brain surface allowed us to compare the spatial distribution of L* with the resection range of the extended lesionectomy.

Results: In each of the six patients who became seizure-free,

maximum values of L* were restricted to recording sites coin- ciding with the area of resection. In contrast, L* maps of most patients who had no benefit from the resection indicated a more widespread extent or the existence of additional, probably autonomous, foci. The mean of Id* values obtained from recording sites outside the area of resection correctly distinguished 13 patients (86.7 %) with respect to seizure outcome.

Conclu,sion,s: Relevant information obtained from long- lasting interictal electrocorticographic recordings can be com- pressed to a single L* map that contributes to a spatial characterization of the primary epileptogenic area. In neocortical lesional epilepsies, L* allows for identification and characterization of epileptogenic activity and thus provides an additional diagnostic tool for presurgical assessment. Key Words: Neo- cortical epilepsy-Primary epileptogenic area-Neuronal complexity loss-Effective correlation dimension-ECoG.

In neocortical lesional epilepsies (NLE) seizure control can be achieved by surgical removal of the lesion if the primary epileptogenic area is included (1-5). Usu- ally, invasive electrophysiological techniques that record the region of seizure origin and interictal spiking define the extent of the area to be resected. In addition to structural imaging techniques such as magnetic resonance imaging (MRI) or cranial computer tomography (CCT), functional techniques such as positron emission tomography or single photon emission computed tomography (6-8) as well as noninvasive recordings of brain electro- magnetic activity (9) are considered the basis to define the primary epileptogenic area. However, in NLE, this area is not necessarily identical with the lesion zone (lo), although both areas are related in the most patients (1 1). Thus, the current standard for the localization and de-

Accepted February 4, 2000. Address correspondence and reprint requests to Guido Widman at

Clinic for Neurology, University Hospital Essen, Hufelandstr. 55, 45 122 Essen, Germany. E-mail: Guido.Widman @uni-essen.de

marcation of the primary epileptogenic area is to record the patient’s spontaneous habitual seizures using chronic electrocorticography (ECoG). This method, however, in- cludes an additional risk for the patient and, depending on the individual seizure frequency, represents a time- consuming and cost-intensive procedure for the presurgical work-up. In addition, the relevance of electro- graphical features as enhanced low frequencies or spike densities used for the interictal identification of the primary epileptogenic area is still matter of discussion (12). In the case of extratemporal neocortical lesions, only 67% of the patients benefit from surgery (1 3) despite the immense presurgical work-up, including unambiguous identification of morphological lesions.

Nonlinear time series analyses (ref. 14 and references therein) have been repeatedly applied to electroencephalographic recordings of epilepsy patients. Extracting nonlinear measures such as the correlation dimension or the largest Lyapunov exponent has yielded promising results in characterizing the epileptic disorder (15-20). Concerning the spatiotemporal dynamics of the epilep-

811

812 G. WIDMAN ET AL.

togenic process, we interpreted temporary transitions from high- to low-dimensional system states, neuronal complexity loss (L*), as representing synchronization phenomena of the activity of neurons involved. These transitions are of high relevance for the lateralization of the primary epileptogenic area in patients with mesial temporal lobe epilepsy (1 8,20). In addition, the neuronal complexity loss proved sensitive for the investigation of anticonvulsant drug effects on the epileptogenic process (21). Moreover, nonlinear measures showed that specific and long-lasting changes in the preictal period were in- dicative of impending seizures (22-25).

However, these findings were mainly based on analyses of recordings in patients with mesial temporal lobe epilepsy. To prove an extended applicability of nonlinear time series analysis beyond this well-defined syndrome, we applied these methods to recordings of patients suffering from NLE of different origin. We specifically examined whether the L* extracted from interictal ECoG recordings allows spatial characterization of the primary epileptogenic area in neocortical lesional epilepsies.

METHODS

Patient data A total of 15 patients (nine women and six men; mean

age, 30 k 10 years; range, 9-46) with a mean age at onset of epilepsy of 15 f 9 years (range, 3-38) were included in the study (Table 1). Each patient was diagnosed as having a pharmaco-resistant lesional neocortical epilepsy of temporal or extratemporal origin. In each case, noninvasive diagnostical methods such as MRI, positron emission tomography, single photon emission computed tomography, or surface EEG could not accomplish the concordant localization of the primary epileptogenic area or its exact demarcation from those regions of the brain whose resection would lead to intolerable neurological

deficits. Thus, the implantation of a subdural 64-contact grid electrode covering the lesion was necessary (following the Bonn protocol for presurgical assessment (26), approved by the local medical ethics committee). In all patients, the grid (intercontact distance, 10 mm) was supplemented by subdural strip electrodes. In five patients, diagnostic findings of the noninvasive phase indicated an involvement of mesial temporal structures, necessitating the implantation of additional intrahippo- campal depth electrodes. In 12 patients, presurgical work-up led to an extended lesionectomy. In the remaining three patients, the latter was restricted because of nearby functionally relevant areas. Postoperatively, all patients have been in the care of our out-patient department for more than 1 year. Informed consent was obtained from all patients.

For evaluation of results, the patients were divided into two groups according to their outcomes (1 3): group A comprised six patients who were seizure-free, corre- sponding to the outcome-class 1A, and group B comprised nine patients in whom the resection did not lead to complete seizure control (classes 2A-4B).

Recording techniques ECoG and stereoelectroencephalograms were re-

corded during the invasive phase of presurgical assessment using an average common reference (18). The sig- nal was bandpass-filtered (0.53-85 Hz; 12 dB/octave) and, after 12-bit A/D conversion, was digitally stored at a sampling frequency of 173 Hz.

Estimation of L* A total of 52 artifact-free ECoG recordings (mean du-

ration, 35 min) of the interictal state from the awake patient (2-7 recordings per patient) were analyzed using a moving-window analysis of an effective correlation dimension DPtf. Details of the methods applied have

TABLE 1. Clinical data ofpatients

Age at Follow-up ID“ Outcome” Age onset (months) Side Region Histology

P-01 IA 30 19 20 Right Parietotemporal Contusion P-02 1A 29 25 25 Left Frontal Astrocytoma WHO I

P-04 1A 21 15 40 Left Temporal Ganglioglioma WHO 1 P-05 IA 43 13 30 Right Parietal Cortical malformation P-06 IA 9 3 17 Left Occipitotemporal Glioneuronal harmatoma P-07 2A 26 17 53 Left Parietotemporal Astrocytoma WHO 2 P-08 2B 18 13 68 Left Temporal Ganglioglioma WHO 1 P-09 2B 42 24 22 Right Temporal Contusion P-I0 3A 28 18 30 Right Frontoparietal Gliosis P-11 4A 32 4 52 Right Frontoparietd Contusion P-12 4A 39 8 73 Left Temporal Gliosis P-13 4A 34 19 35 Left Frontal Cystic malformation P-14 4A 24 4 41 Left Parietal Ganglioglioma WHO 1 P-15 4B 30 9 30 Left Parietal Oligodendroglioma WHO 2

P-03 1A 46 38 28 Left Frontal Astrocytoma WHO 2-3

a ID, identification number of each patient. ’ Postoperative seizure status according to (13).

&ikpsiu, Vol. 41, No. 7, 2000

L* MAPS IN NEOCORTICAL EPILEPSIES 813

been described elsewhere (18,23). Briefly, data sets were segmented into consecutive and normalized epochs of 30 sec duration (N = 5,190 data points) for which quasis- tationarity may be assumed (27). After digitally low- pass-filtering the data (cutoff frequency, 40 Hz; 4th order Butterworth characteristic), correlation sums C, (r) and their local derivatives C', (r) were calculated for embedding dimensions m = 1 to 30 using an optimized Grassberger-Procaccia algorithm (28,29). The range of possible values for the hypersphere radius r was selected to match the resolution of the A/D converter. The time

FIG. 1. A: Representative examples of dimension profiles extracted from an interictal ECoG segment (duration, 20 min). Upper rows are recording sites (82 and C2) within the primary epileptogenic area; lower rows are remote recording sites (B8 and C8). Pro- files were smoothed for better visualization using a three-point moving average. B: L* value for each recording site of the 64- contact subdural grid electrode. In this example, maximum values are confined to recording sites co- inciding with the area of resection (gray bars). C: Projection of electrode contacts (obtained from a postimplant lateral x-ray exposure) onto a postoperatively obtained sagittal MRI scan. The area of resection is marked black and encircled white.

5

0 10

5

Deff 0 * 10

5

0 10

5

0

delay for the phase space reconstruction (30) was set to one sampling point, and a Theiler cutoff of five sampling points was used to limit autocorrelative effects (31). A reliable estimation of D2eff requires the existence of a "scaling region" of a sufficient length. However, in the analysis of electroencephalographic data, this require- ment is hardly fulfilled in the strict sense. We therefore defined a "quasiscaling" that had to meet the following weaker requirements: starting from the one-dimensional embedding (m = l), the upper bound r, of the quasiscaling was defined as the r value at which C' ,(r) exceeds

t -1

1 , , , , 1 / , / / 1 1 , l " " 1 " ' ~ I " "

0 5 10 15 20 t [min]

Epilepsiu, Vol. 41, NO. 7, 2000


0.95. The lower bound rl was reached when CZ5(r) > 1.05 C’25(ru) for decreasing r. D;If was assigned the average of C’zs (r) values within the range [rl, rJ, pro- vided the interval [r,, r,] was long enough (at least one octave, i.e., at least five consecutive C’2s(r) values, con- tributed to the average) and DZCff was lower than a theo- retical resolution limit (32) of 210g1,,(N) = 7.43. In all other cases, Dgff was set to a fixed value of D,, = 10. The time-independent measure L* was extracted from the resulting Dzeff profiles (Fig. 1A). L* represents the area between the DZcff profile and the D,, line, normalized with respect to the observation time (18). For each recording contact of the grid, means of L* values (F) were obtained by averaging results obtained for all ECoG data sets under consideration. Because absolute values of DPrs and L* depend on recording, filtering, and calculation parameters, only relative differences can be investigated. The resulting spatial distribution (L* maps) was normalized (i.e., mapped to to ,]]) , thus al- lowing for comparison of results obtained from different patients (Fig. lB).

Matching L* maps and area resected To relate the spatial distribution of to the area

resected, the location of grid electrodes was determined with respect to anatomical landmarks. A lateral x-ray exposure of the skull was performed during presurgical assessment with the patient in a supine position. The film was placed ipsilateral to the side of the implanted grid, and the focus film distance was set to 1 50 cm to approxi- mate a parallel path of x-rays. To identify the area of resection, an MRI was performed 5 6 months after the lesionectomy (sagittal T1 -weighted spin echo images;

slice thickness 5 mm; interslice gap 0.5 mm). The x-ray exposure depicting the grid was matched with the medial sagittal MRI slice using the boundary of the calvarium, the sella turcica, and the odontoid process as anatomical landmarks. The hemisphere onto which the grid had been implanted was segmented from cerebrospinal fluid and scar tissue in all sagittal MRI sections using a semiau- tomatic algorithm. Resulting brain slices were superim- posed and projected on the medial MRT slice. Finally, markers of the electrode contacts were projected onto the surface of the brain (Fig. IC).

Electrode contacts were split into two groups according to their location with respect to the area resected. On average, 19 contacts per patient (+lo; range, 7-32) were identified within or on the boundary of the area resected (C,,) in group A; in group B, these values amounted to 15 per patient (27; range, 6-25). The number of contacts outside the area resected (Gout) was 44 ( k l l ; range, 32- 57) for group A and 48 (+8; range, 38-58) for group B. The differences between patient groups were nonsignif- icant for both Ci, and C<,ut (Kolmogorov-Smirnov Z test).

Statistical analyses were performed on Lati,, and L*oul values, representing the mean of F values within and outside the area resected, respectively.

RESULTS

L* maps in relation to the area resected Fig. 2 depicts typical examples of spatial F distribu-

tion. In all group A patients, maximum values and those above the 80% quantile were restricted to recording sites covering the area of resection (Fig. 2A). In four group B patients, maximum values were also inside

FIG. 2. L*maps projected onto the brainsurface for patients 4 (A), 10 (B), and 13 (C) (Table 1) with a 50% transparency to preserve underlying brain structures. Normalized L” values were encoded using a color scale as shown below. Pixels between contacts were interpolated using a two-dimensional second-order spline function (33) for better visualization of results.

Epikpsia, Vol. 41, No. 7, 2000


the area of resection. In two of these patients, values above the 80% quantile were also found outside the area of resection; whereas in two patients, maximum values were found at recording sites at the edge of the grid, suggesting an insufficient coverage of the primary epileptogenic area. In the five remaining group B patients,

maximum values were found at recording sites away from the area of resection (Fig. 2B, C).

Fig. 3 depicts the distribution of L* and the resected area in relation to the grid electrode for each patient. For comparison, a summary of ECoG-related findings of the presurgical work-up, including electrocorticographical

FIG. 3. L* maps, extent of resection, and ECoG-related findings of the presurgical work-up for all patients. Upper left square IS the distribution of L*; upper right square is the resected area (black boxes) in relation to the grid electrode. Gray boxes indicate eloquent brain areas as defined by electrical stimulation. Brown boxes indicate an overlap of these two areas. Lower two squares are a summary of ECoG activity obtained during presurgical work. Left: ictal state; black boxes, seizure onset; gray boxes, seizure spread, excluding secondary generalized seizures; brown boxes, overlap of the two activity areas. Right: interictal state; black boxes, epilepsy specific focal activity; gray boxes, unspecific focal activity; brown boxes, overlap of the two activity areas. In patient 1 and 14, no seizures occurred during presurgical assessment. It should be noted that these plots present findings of ictal and interictal ECoG activity only. Other evaluation techniques (e.g., seizure semiology or imaging) contribute as well to the definition of the area of resection (26).

Epllr.psiu, V d . 41, N o 7, 2000


mapping of eloquent areas and ictal and interictal ECoG activity obtained by visual analysis, is shown.

Ally interpretation of these findings should account for the location of high values in relation to functionally relevant areas. In two group B patients (patients 14 and IS), electrical cortical stimulation determined the functional constraints on the resection. In these patients, the primary epileptogenic area was found close to either mo- tor cortex or Broca’s area. Evaluation of maps of the seven remaining group B patients showed high values in functionally relevant areas in four of the patients (patients 7, 12, 13, and 14).

Estimating the spatial extent of the primary epileptogenic area

Multivariate analysis of variance showed significant effects of seizure outcome on mean values obtained from inside or outside the area resected (p < 0.01). Post hoc univariate analysis of variance using L*in and L*,,,, values as dependent variables yielded significant differences between the two outcome groups for L*otri values (Fig. 4). Discriminant analysis using L*o,Lt values allowed all but two patients (patient 8 and 9) to be distinguished correctly. In one of these cases, the highest L* values were restricted to few contacts at the edge of the grid (Fig. 2).

DISCUSSION

It is likely that the primary epileptogenic area in NLE is closely related to a lesion (11). The demarcation of epileptogenic tissue can be simple, as in the case of solid, benign tumors that are well defined by structural imaging techniques. Other lesions, e.g., dysplasias, may be more diffuse, and imaging techniques can fail to detect such alterations. Thus, identification and demarcation of a brain area that generates seizures requires a close spatial sampling of ictal and interictal epileptiform events.

Resection of a lesion along with the surrounding brain

1 .o

0.8 a e 0.6

2 0.4 E

0.2

0.0

0, C

**

in out recording site

FIG. 4. Group means and standard deviations of the neuronal complexity loss t*in and L*,,, at recording sites within and outside the resection range for seizure outcome groups A (filled bars) and B (empty bars). Asterisks denote significant intergroup differences (p < 0.005, Mann-Whitney U test).

tissue may not guarantee complete seizure control (1 1). Postoperative persistence of seizures may occur for sev- eral reasons. Known problems caused by the surgery include epileptogenicity of the scar or a restricted extent of resection (e.g., to avoid neurological deficits). Sei- zures may also persist due to specific properties of the epileptic process, such as secondary epileptogenesis that may generate additional foci (34).

Thus, although different methodologies are available to localize the primary epileptogenic area and demarcate it from functionally relevant areas (3) , an unequivocal characterization of the spatial extent of the epileptogenic process requires additional information. Despite the few patients investigated so far, our results suggest that relevant information can be obtained from nonlinear time series analysis of the interictally recorded electrocortico- gram in NLE. By taking advantage of surgical outcome as the most important validation criterion, we showed that circumscribed areas of maximum L* matched the resected area in patients who benefited from surgery. In contrast, multiple and widespread areas of increased L* were found in all but two patients who continue suffering from seizures despite an extended lesionectomy. These areas were predominantly located apart from the area resected and, in some cases, close to or within functionally relevant area. This suggests the existence of additional, possibly autonomous foci that were either not identified or not sufficiently considered by conventional methods used in the presurgical work-up (26). It is ob- vious that the presented method cannot explain persistence of seizures in all cases, especially not in those cases with an insufficient coverage of epileptogenic brain areas. In summary, our results show that the nonlinear measure neuronal complexity loss provides additional information for a spatial characterization of the primary epileptogenic area in neocortical lesional epilepsy, even during the interictal state. Nonlinear time series analysis can thus contribute to an improvement of the presurgical assessment. Future studies should include more complex cases (e.g., nonlesional neocortical epilepsy or involvement of mesial temporal structures in neocortical epilepsies) as well as investigations of the applicability of the proposed method in acute electrocorticography .

Acknowledgment: We thank our neurosurgical colleagues J. Schramm, M.D., J. Zentner, M.D., D. Van Roost, M.D., and E. Behrens, M.D., who implanted the intracranial electrodes and performed the lesionectomies, as well as M. Kurthen, M.D. and W. Burr, Ph.D., for helpful comments. This study was sup- ported by the Deutsche Forschungsgemeinschaft (grant no. EL 122/3-2).

REFERENCES I . Salanova V, van Ness PC, Andermann F. Frontal, parietal, and

occipital epilepsies. In: Wyllie E, ed. The treafment of epilepsy: principle& and practice. 2nd edition. Baltimore: Williams & Wilkins, 1996:423-3 1.

Epilepspsia, Vol. 41, No. 7, 2000


2.

3.

4.

5.

6.

7

8 9

10

1 1

12

13

14

15

16

Zentner J, Hufnagel A, Ostertun B, et al. Surgical treatment of extratemporal epilepsy: clinical, radiological, and histopathologi- cal findings in 60 patients. Epilepsia 199637: 1072-80. Comair YC, Choi HY, van Ness PC. Neocortical resections. In: Engel 1 Jr, Pedley TA, eds. Epilepsy: a comprehensive textbook. Philadelphia: Lippincott-Raven, 1997: 18 19-28. Fried I, Cascino GD. Lesionectomy. In: Engel J Jr, Pedley TA, eds. Epilepsy: a comprehensive textbook. Philadelphia: Lippincott- Raven, 1997:1841-50. Tran TA, Spencer SS, Spencer DD. Epilepsy: Medical and surgical outcome. In: Swash M, ed. Outcomes in neurological and neurosurgical disorders. New York: Cambridge University Press, 1998: 407-40. Hajek M, Antonini A, Leenders KL, Wieser HG. Mesiobasal ver- sus lateral temporal lobe epilepsy: metabolic differences in the temporal lobe shown by interictal IRF-FDG positron emission tomography. Neurology 1993;43:79-86. Wieser HG. PET and SPECT in epilepsy. Eur Neurol 1994;34 (suppl 1):58-62. Duncan JS. Imaging and epilepsy. Brain 1997;120:339-77. Nakasato N, Levesque MF, Barth DS, Baumgartner C, Rogers RL, Sutherling WW. Comparisons of MEG, EEG, and ECoG source localization in neocortical partial epilepsy in humans. Electr-oen- cephalogr Clin Neurophy,siol 1994;91:171-8. Luders HO, Awad 1. Conceptual considerations. In: Liiders HO, ed. Epilep.sy surgery. New York: Raven Press, 1991 :5 1-62, Wieser HC. Epilepsy surgery. Baillieres Clin Neurol 1996;5:849- 7 c

ronal complexity loss. Electroencephalogr Clin Neurophysiol 1995;95:108-17.

19. Casdagli MC, Iasemidis LD, Savit RS, Gilmore RL, Roper SN, Sackellares JC. Non-linearity in invasive EEG recordings from patients with temporal lobe epilepsy. Electroencephnlugr Clin Neurophysiol 1997; 102:98- 105.

20. Weber 9, Lehnertz K, Elger CE, Wieser HG. Neuronal complexity loss in interictal EEG recorded with foramen ovale electrodes pre- dicts side of primary epileptogenic area in temporal lobe epilepsy: a replication study. Epilepsia 1998;39:922-7.

21. Lehnertz K, Elger CE. Neuronal complexity loss in temporal lobe epilepsy: effects of carbamazepine on the dynamics of the epileptogenic focus. Electroencephalogr Clin Neurophysiol 1997; 103: 376-80.

22. Elger CE, Lehnertz K. Seizure prediction by noii-linear time series analysis of brain electrical activity. Eur J Neurosci 1998; 10:786-9.

23. Lehnertz K, Elger CE. Can epileptic seizures be predicted? Evi- dence from nonlinear time series analyses of brain electrical activity. Phys Rev Lett 1998;80:5019-23.

24. Martinerie J, Adam C, Le Van Quyen M, et al. Epileptic seizures can be anticipated by non-linear analysis, Nut Med 1998;4: 1173-6.

25. Moser HR, Weber B, Wieser HG, Meier PF. Electroencephalo- grams in epilepsy: analysis and seizure prediction within thc framework of Lyapunov theory. Physica D 2000; In press.

26. Presurgical evaluation protocols. In: Engel J Jr, ed. Surgical freat- ment of the epilepsies. 2nd edition. New York Raven Press Ltd., I993:740-2.

I J. 27. Lopes da Silva FH. EEG analysis: Theory and practice. In: Nie- dermayer E, Lopes da Silva FH, eds. Electroencephalography, basic principles, clinical applications and related fields. Balti- more, MD: Urban and Schwarzenberg, 19872371.

28. Grassberger P, Procaccia I. Measuring the strangeness of strange attractors. Physica D 1983;9: 189-208.

29. Widman G, Lehnertz K, Jansen P, Meyer W, Burr W, Eker CE. A fast general purpose algorithm for the computation of auto- and cross-correlation integrals from single channel data. Physica D 1998;121:65-74.

30. Takens F. Detecting strange attractors in turbulence. In: Rand DA, Young LS, eds. Lecture notes on mathematics. New York: Springer,l981:366-81.

31. Theiler J. Spurious dimension from correlation algorithms applied to limited time series data. Phys Rev A 1986;34:2427-32.

32. Ruelle D. Deterministic chaos: the science and the fiction. Proc R

Tran TA, Spencer SS, Javidan M, Pacia S, Marks D, Spencer DD. Significance of spikes recorded on intraoperative electrocorticography in patients with brain tumor and epilepsy. Epilepsia 1997; 38:1132-9. Engel Jr, Van Ness pC, Rasmussen TB, ojemann LM, Outcome with respect to epileptic seizures. In: Engel J Jr, ed. Surgical treat-

?[ the eI,ilepsies, 2nd edition, N~~ York: R~~~~ press Ltd, 1993509-22. Kano H, Schreiber T, eds. Nonlinear time series analysis. Cam- bridge: Cambridge University Press, 1997. Iasemidis LD, sackellares JC, zaveri HP, wi1liams WJ, Phase space topography and the Lyapunov exponent of electrocorticograms in partial seizures. Brain Topogr 1990;2: 187-201, pijn jp, van N~~~~~~ J, ~~~~t A, L~~~~ da silva FH, Chaos or noise in EEG signals: dependence on state and brain site. Electro- encenhulowr Clin Neuroohvsial 199 1:79:37 1-81.

17. Pi,in.JP, fe l i s DN, van be; Heyden MJ, DeGoede J, van Veelen CW, Lopes da Silva FH. Nonlinear dynamics of epileptic seizures on basis of intracranial EEG recordings. Brain Topogr 1997;9: 249-70. logr Clin Neurophysiol 1987;66:75-81.

18. Lehnertz K, Elger CE. Spatiotemporal dynamics of the primary epileptogenic area in temporal lobe epilepsy characterized by neu-

Sac Lond 1990;A427:241-8. 33. Perrin F, Pernier J, Bertrand 0, Giard MM, Echallier JF. Mapping

of scalp potentials by surface spline interpolation. Electroencepha-

34. Morrell F. Secondary epileptogenesis in man. Arch Neurol 1985; 42: 3 18-35.

Epilepsia, Vol. 41, No. 7, 2000

complexidade e epilepsia

Documents