Bilateral oscillations for lateralized spikes in benign rolandic epilepsy

Bilateral oscillations for lateralized spikes in benign rolandic epilepsy

Epilepsy Research 69 (2006) 45–52 Bilateral oscillations for lateralized spikes in benign rolandic epilepsy Yung-Yang Lin a,b,c,e,f,∗ , Fu-Jung Hsiao...

396KB Sizes 0 Downloads 132 Views

Epilepsy Research 69 (2006) 45–52

Bilateral oscillations for lateralized spikes in benign rolandic epilepsy Yung-Yang Lin a,b,c,e,f,∗ , Fu-Jung Hsiao a,e , Kai-Ping Chang d,g , Zin-An Wu b,f , Low-Tone Ho a,e a

Institute of Physiology, National Yang-Ming University, Taipei, Taiwan Department of Neurology, National Yang-Ming University, Taipei, Taiwan c Institute of Clinical Medicine, National Yang-Ming University, Taipei, Taiwan d Department of Pediatrics, National Yang-Ming University, Taipei, Taiwan Department of Medical Research and Education, Taipei Veterans General Hospital, Taipei, Taiwan f Neurological Institute, Taipei Veterans General Hospital, Taipei, Taiwan g Department of Pediatrics, Taipei Veterans General Hospital, Taipei, Taiwan b

e

Received 2 November 2005; received in revised form 7 December 2005; accepted 8 December 2005 Available online 28 February 2006

Abstract Purposes: To elucidate the oscillatory dynamics with respect to interictal spike occurrence in benign rolandic epilepsy (BRE). Methods: Using a whole-scalp magnetoencephalography (MEG), we recorded scalp EEG and MEG signals in 10 BRE patients (age 8–12 years) and visually identified unilateral interictal spikes that were simultaneously present on both EEG and MEG channels. We obtained the peak timing of individual spike complex based on MEG single-dipole modeling, and then applied wavelet transform to analyze the time–frequency components of corresponding MEG signals with respect to spike occurrence. Results: In the hemisphere with time-domain spike waveforms, we identified a clear increase of 0.5–40 Hz activity around the spike peak, most prominent at alpha band (8–13 Hz). Notably, at the approximate timing we also observed an increase in 0.5–25 Hz oscillations over the homotopic area in the other hemisphere where no spike signals were found. Conclusions: Our results indicate bilateral increases in 0.5–25 Hz oscillations during unilateral spike formation in BRE patients. By using wavelet transform analysis, one could be able to detect some irritative feature that would in visual analysis remain undetected. © 2006 Elsevier B.V. All rights reserved. Keywords: Oscillatory dynamics; Benign rolandic epilepsy; Wavelet transform; Magnetoencephalography; Time–frequency representation; Interictal spikes



Corresponding author. Present address: Department of Medical Research and Education, Taipei Veterans General Hospital, No. 201, Sec. 2, Shih-Pai Rd., Taipei 112, Taiwan. Tel.: +886 2 2873 2902; fax: +886 2 2871 3010. E-mail address: [email protected] (Y.-Y. Lin). 0920-1211/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.eplepsyres.2005.12.006

1. Introduction Benign rolandic epilepsy (BRE) is regarded as a focal epilepsy syndrome (Lombroso, 1967). Interictal spikes, typically localized around the central sulcus,

46

Y.-Y. Lin et al. / Epilepsy Research 69 (2006) 45–52

may be observed in one or both hemispheres (Lerman and Kivity, 1975; Lin et al., 2003a,b). Recent studies show evidence of inter-hemispheric source shifting (Kellaway, 2000; Lin et al., 2003a; Huiskamp et al., 2004; Ishitobi et al., 2005). However, the exact neuronal mechanisms underlying the cortical excitability remain not clear. Wavelet transform (WT) analysis offers feasible frequency and temporal evaluations for non-stationary signals (Akin, 2002), and may reveal some dynamic information not detected by time-domain-based evaluation. To explore the oscillatory dynamics of rolandic discharges, we identified interictal spikes from simultaneous MEG and scalp EEG recordings in 10 patients and analyzed the oscillatory characteristics by WT.

2. Methods 2.1. Patients We studied 10 children (3 boys, 7 girls; age 8–12 years) who were diagnosed as benign rolandic epilepsy according to infrequent motor seizures, normal neurodevelopment and evidence of centrotemporal spikes on scalp EEG recordings (Lombroso, 1967). All patients and their parents gave informed consent to the experimental procedures. All patients were normal at the neurological examination and had no brain lesions on magnetic resonance (MR) images. 2.2. MEG and scalp EEG recordings After placing scalp EEG electrodes according to the International 10–20 system, we simultaneously recorded MEG and EEG signals for 30–40 min for each patient with a whole-scalp neuromagnetometer (VectorviewTM , Elekta Neuromag, Helsinki, Finland). The raw data were bandpass filtered between 0.1 and 130 Hz, sampled at a digitization rate of 400 Hz, and stored in magnetic optical disks for off-line analysis. 2.3. Identification of MEG epochs with unilateral spikes We identified unilateral spikes by visually evaluating EEG (A1–A2 referential and double-banana bipolar montages) and MEG signals, and then enrolled 600ms MEG epochs containing unilateral single spikes

for further evaluation. We excluded those epochs with another neighboring spikes, with background artifacts, or with bilateral spikes. The 600-ms MEG epoch centered by one spike complex was then evaluated by equivalent current dipole (ECD) model. Around the main spike peak, we identified the ECDs with adequate goodness-of-fit values by a least-squares search using subsets of 40–60 channels around the maximum signals. These calculations resulted in the location, orientation and strength of the ECD in a spherical conductor model (H¨am¨al¨ainen et al., 1993) on the basis of the brain MR images acquired with a 3-T Bruker Medspec300 Scanner (Germany). For each MEG epoch, we accepted one single ECD that best explained the main spike signals. Based on the ECD waveform as a function of time, we identified the peak timing of the main spike (Minami et al., 1996; Lin et al., 2003a,b). For a more friendly communication, we re-selected the 600-ms epoch with the peak timing of the MEG spike source at the center (0 ms at time scale). In each patient, we obtained 10 MEG epochs with unilateral spikes for subsequent WT analysis. 2.4. Wavelet analyses Time–frequency analyses of the 600-ms MEG epochs containing unilateral spikes were performed by a continuous WT. We used the Morlet wavelet (Kronland-Martinet et al., 1987), which is a function of time t and frequency f0 , defined as:  2 −t w(t, f0 ) = A exp exp(i2πf0 t). 2σt2 The width of the wavelet (m = f0 /σ f ) was chosen to be 7 (Tallon-Baudry et al., 1996, 1998; Lachaux et al., 1999; Rodriguez et al., 1999; Jensen et al., 2002); where A = 1/2 1/(2πσt2 ) , σ t = 1/(2πσ f ). The time-varying power of the neuromagnetic signals in a frequency band around f0 is the squared norm of the result of the convolution of the complex wavelet w(t, f0 ) with the signal s(t): E(t,f0 ) = [w(t, f0 )×s(t)]2 . We used a set of wavelets with frequencies, f0 , covering the 0.5–40 Hz range at intervals of 1 Hz. 2.5. Data presentations and statistics For each MEG channel, the E(t,f0 ) was averaged across individual bands of 0.5–4, 4–8, 8–13, 13–25

Y.-Y. Lin et al. / Epilepsy Research 69 (2006) 45–52

and 25–40 Hz to provide the time-varying measures of delta, theta, alpha, beta and gamma activities, respectively. In the spike hemisphere, the channel with most prominent spectral power around 0 ms was selected. We calculated the mean power around the spike period (i.e. from −25 to +25 ms) and baseline period (i.e. from −300 to −250 ms). For the signal of the homotopic channel in the non-spike hemisphere, we also calculated mean power in corresponding time periods for each frequency band. To find the most prominent oscillation during rolandic spiking, we compared the mean peak power between different frequency bands by one-way ANOVA with frequency bands (delta, theta, alpha, beta and gamma) as a within-subject factor, and used Bonferroni test for post-hoc tests. To further show the spatial dynamics of power change for the main frequency band, we calculated the

47

relative power overtime with respect to the mean power throughout the 600-ms epoch. We also made color plots on Neuromag helmets to show the spatial distribution of oscillatory activity by using the topohelmet function of 4-D toolbox. Across 10 spike epochs in each patient, we used a paired t-test to assess the power difference between the spike and the baseline periods, and then displayed the statistical results by mapping the t-values over the head topography to show the spatial dynamics of the main oscillation with respect to the occurrence of rolandic spikes. p-values < 0.05 were considered significant.

3. Results Fig. 1 shows the time-domain and frequencydomain features of interictal spikes from Patient 1.

Fig. 1. (A) Simultaneous MEG and EEG signals, low-pass filtered at 40 Hz, from Patient 1 with interictal spikes (black arrows). The eight MEG channels show signals of planar gradiometers: four from the left (LH) and four from the right hemisphere (RH). ECG, electrocardiogram. (B) Topographic distribution of the unilateral MEG spikes. The head is viewed from the top and each tracing illustrates the signal recorded by one gradiometer. MR images in the bottom show the spike location (white dot). R, right; L, left; A, anterior; P, posterior. (C) Time–frequency representations between 0.5 and 40 Hz of the 600-ms MEG epochs for all gradiometers. The amplitude is color-coded; small amplitudes are indicated with blue and large with red. (D) Spatiotemporal pattern of alpha power change with respect to the occurrence of a rolandic spike in the left hemisphere.

48

Y.-Y. Lin et al. / Epilepsy Research 69 (2006) 45–52

Fig. 2. Spatial distribution of alpha activity with respect to the main peak (defined as 0 in time) of unilateral spikes in the left (Patient 2) and the right hemisphere (Patient 9). Lateral views of the head are shown with the nose pointing to the left or the right. The alpha power is color-coded; large power is denoted with red and small with blue.

Although spike waveforms were identified only in the left hemisphere (see Fig. 1A and B), the concomitant oscillatory enhancement occurred homotopically in both hemispheres (see Fig. 1C). The relative alpha power in all channels as a function of time further disclosed the bilateral power increase around 0 ms (see Fig. 1D). Table 1 shows the mean (±S.E.M.) power across 10 patients of various oscillations around the baseline and the spike peak. In the spike hemisphere, the epileptic spiking was associated with a clear increase in the oscillations ranging from 0.5 to 40 Hz. In the non-

spike hemisphere, the 0.5–25 Hz oscillation was also increased around the peak timing of the contralateral spike. Of these bands, alpha was the most prominent component (p < 0.05 compared with other frequency bands). Fig. 2 demonstrates the spatial distribution of alpha activity overtime with respect to the peak (coded as 0 ms) of time-domain spike in Patients 2 and 9 who had corresponding unilateral spikes in the left and right hemisphere, respectively. Around 50 ms before and after the spike peak, the increased alpha was observed over bilateral frontotemporal regions, with a larger

Peak

3481 ± 1256* 402 ± 46

Baseline

492 ± 69 361 ± 38

4. Discussion

25540 ± 955 ± 76*** 715 ± 81 599 ± 61

Peak Baseline

1348 ± 157 55010 ± 892 ± 151 2202 ± 310***

∗∗∗

p < 0.05 compared with the corresponding baseline data. p < 0.005 compared with the corresponding baseline data. p < 0.001 compared with the corresponding baseline data. ∗

5120 ± 860 7258 ± 664 ± 98 777 ± 93** Spike side Non-spike side

∗∗

4181 ± 759 31695 ± 935 ± 126 1586 ± 178*

Baseline Peak

1391**

Baseline

Peak

6274**

Baseline

Peak

9043***

Beta (13–25 Hz) Alpha (8–13 Hz) Theta (4–8 Hz) Delta (0.5–4 Hz)

Table 1 Mean (±S.E.M.) power (fT2 /cm2 ) across 10 subjects of various oscillatory activities around the spike peak and pre-spike baseline

49

power on the spike hemisphere than the contralateral side. Fig. 3 further shows the bi-hemispheric alpha increase during the occurrence of unilateral spikes in the 10 patients.

3200***

Gamma (25–40 Hz)

Y.-Y. Lin et al. / Epilepsy Research 69 (2006) 45–52

Fast Fourier transform has been used to study the dominant frequencies of EEG signals (Haykin et al., 1996; Muthuswamy and Thakor, 1998). However, that method did not provided with sufficient temporal information. Using WT analysis, our present work shows clearly increased 0.5–40 Hz activities during rolandic spiking, with maximal power at alpha range. The timing for peaking alpha power was temporally compatible with the timing for the main spike peak. Our results agreed with earlier observation that wavelet analysis seems ideally suited to measure the time of occurrence and the location of small-scale transient events such as focal epileptogenic spikes (Samar et al., 1999). Wavelet analysis has the advantage of improving time or space resolution as the duration of an event in the time-domain or the topographic distance over the scalp covered by an event decreases (Samar et al., 1999). Several studies have employed the wavelet analysis in EEG spike identification (Kalayci et al., 1994; Schiff et al., 1994; Tang and Ishii, 1993). Notably, our results further showed evidence of bilateral oscillations for the lateralized time-domain spikes, suggesting a synchronized activity in a network of bilateral rolandic neurons. We suggest WT an appropriate tool for capturing the intricate, frequency-based dynamics of rolandic spikes. Thus, that by this means you could be able to detect irritative activity that would in visual analysis remain undetected. Along with unilateral spike formation, the increase in 0.5–25 Hz activity was almost simultaneously found in homologous areas of both hemispheres. The relatively focal distribution in each hemisphere and the lack of gradual power changes from one hemisphere to the other do not suggest a neighboring propagation. In contrast, Vanhatalo et al. (2004) have reported that the 0.02–0.2 Hz oscillations in widespread cortical regions were strongly synchronized with K-complex and interictal epileptic events. They suggest that the generalized infraslow oscillations represent a cyclic

50

Y.-Y. Lin et al. / Epilepsy Research 69 (2006) 45–52

Fig. 3. The t-values color maps showing the results of a statistical comparison of alpha power ratios for the spike period vs. the baseline period. Time-domain spike waveforms are found on the left hemisphere for Patients 1–5 and the right hemisphere for Patients 6–10. According to the statistically color-coded scale, yellow (t-value = 2.26) and above indicate significant power increase compared with the baseline (p < 0.05).

Y.-Y. Lin et al. / Epilepsy Research 69 (2006) 45–52

modulation of cortical gross excitability, predisposing brain to epileptic activity (Vanhatalo et al., 2004). However, our present data seemed not sufficient to determine whether the 0.5–40 Hz oscillations would reflect a general state for predisposing to spike occurrence. BRE appears to be a maturational or developmental abnormality. Based on the “ascending sequential maturation” theory in mammalian neocortical organization (Kawaguchi et al., 1983), the developmental processes involve a sequential synaptogenesis proceeding from the deep layer to the superficial layer. The maturation disturbance in the thalamocortical development process may underlie the hyper-excitability of peri-central cortices (Kawaguchi et al., 1983). In contrast to the conventional view of focal epileptogenicity, our present data suggest that this developmental disorder may actually cause bilateral cortical hyperexcitability even for unilateral spikes in time-domain presentations. This idea may agree with earlier observations of bilateral independent spikes (Lombroso, 1967; Kellaway, 2000; Lin et al., 2003a) or even generalized spike-wave complexes in BRE (Lerman and Kivity, 1975). Moreover, we suppose a possible involvement of thalamo-cortical circuits in the concomitant increase in alpha oscillation over both hemispheres. Earlier observation of clear correlation between interictal spikes and sleep spindling also suggests a role of thalamocortical connections in the expression of the complex rolandic discharges (Kellaway, 2000; Nobili et al., 1999; Huguenard, 2000). Acknowledgments This study was supported by research grants from Taipei Veterans General Hospital (VGH94-323, V95C1-043) and National Science Council (NSC-932341-B-075-086, NSC-94-2314-B-010-065), Taipei, Taiwan. We highly appreciated the comments from anonymous reviewers that strengthened our present work. References Akin, M., 2002. Comparison of wavelet transform and FFT methods in the analysis of EEG signals. J. Med. Syst. 26, 241–247. H¨am¨al¨ainen, M., Hari, R., Ilmoniemi, R.J., Knuutila, J., Lounasmaa, O.V., 1993. Magnetoencephalography—theory, instrumentation,

51

and applications to noninvasive studies of the working human brain. Rev. Mod. Phys. 65, 413–497. Haykin, S., Racine, R., Xu, Y., Chapman, C., 1996. Monitoring neuronal oscillations and signal transmissions between cortical regions using time–frequency analysis of electroencephalographic activity. Proc. IEEE 84, 1295–1301. Huguenard, J.R., 2000. Circuit mechanisms of spike-wave discharge: are there similar underpinnings for centrotemporal spikes? Epilepsia 41, 1076–1077. Huiskamp, G., van Der Meij, W., van Huffelen, A., van Nieuwenhuizen, O., 2004. High resolution spatio-temporal EEG-MEG analysis of rolandic spikes. J. Clin. Neurophysiol. 21, 84–95. Ishitobi, M., Nakasato, N., Yamamoto, K., Iinuma, K., 2005. Opercular to interhemispheric source distribution of benign rolandic spikes of childhood. Neuroimage 25, 417–423. Jensen, O., Gelfand, J., Kounios, J., Lisman, J.E., 2002. Oscillations in the alpha band (9–12 Hz) increase with memory load during retention in a short-term memory task. Cereb. Cortex 12, 877–882. Kalayci, T., Ozdamar, O., Erdol, N., 1994. The use of wavelet transform as a preprocessor for the neural network detection of EEG spikes. In: Proceedings of the IEEE Southeastcon’94, pp. 1–3. Kawaguchi, S., Samejima, A., Yamamoto, T., 1983. Postnatal development of the cerebello-cerebral projection in kittens. J. Physiol. 343, 215–232. Kellaway, P., 2000. The electroencephalographic features of benign centrotemporal (rolandic) epilepsy of childhood. Epilepsia 41, 1053–1056. Kronland-Martinet, R., Morlet, J., Grossmann, A., 1987. Analysis of sound patterns through wavelet transforms. Int. J. Patt. Recognit. Artiff. Intell. 1, 273–302. Lachaux, J.P., Rodriguez, E., Martinerie, J., Varela, F.J., 1999. Measuring phase synchrony in brain signals. Hum. Brain Mapp. 8, 194–208. Lerman, P., Kivity, S., 1975. Benign focal epilepsy of childhood. A follow-up study of 100 recovered patients. Arch. Neurol. 32, 261–264. Lin, Y.Y., Chang, K.P., Hsieh, J.C., Yeh, T.C., Yu, H.Y., Kwan, S.Y., Yen, D.J., Yiu, C.H., Hari, R., 2003a. Magnetoencephalographic analysis of bilaterally synchronous discharges in benign rolandic epilepsy of childhood. Seizure 12, 448–455. Lin, Y.Y., Shih, Y.H., Chang, K.P., Lee, W.T., Yu, H.Y., Hsieh, J.C., Yeh, T.C., Wu, Z.A., Ho, L.T., 2003b. MEG localization of rolandic spikes with respect to SI and SII cortices in benign rolandic epilepsy. Neuroimage 20, 2051–2061. Lombroso, C.T., 1967. Sylvian seizures and midtemporal spike foci in children. Arch. Neurol. 17, 52–59. Minami, T., Gondo, K., Yamamoto, T., Yanai, S., Tasaki, K., Ueda, K., 1996. Magnetoencephalographic analysis of rolandic discharges in benign childhood epilepsy. Ann. Neurol. 39, 326–334. Muthuswamy, J., Thakor, N.V., 1998. Spectral analysis methods for neurological signals. J. Neurosci. Methods 83, 1–14. Nobili, L., Ferrillo, F., Baglietto, M.G., Beelke, M., De Carli, F., De Negri, E., Schiavi, G., Rosadini, G., De Negri, M., 1999. Relationship of sleep interictal epileptiform discharges to sigma activity (12–16 Hz) in benign epilepsy of childhood with rolandic spikes. Clin. Neurophysiol. 110, 39–46.

52

Y.-Y. Lin et al. / Epilepsy Research 69 (2006) 45–52

Rodriguez, E., George, N., Lachaux, J.P., Martinerie, J., Renault, B., Varela, F.J., 1999. Perception’s shadow: long-distance synchronization of human brain activity. Nature 397, 430–433. Samar, V.J., Bopardikar, A., Rao, R., Swartz, K., 1999. Wavelet analysis of neuroelectric waveforms: a conceptual tutorial. Brain Lang. 66, 7–60. Schiff, S.J., Heller, J., Weinstein, S.L., Milton, J., 1994. Wavelet transforms and surrogate data for electroencephalographic spike and seizure detection. Opt. Eng. 33, 2162–2169. Tang, Z.W., Ishii, N., 1993. The recognition system with two channels at different resolution for detecting spike in Human’s EEG. IEICE Trans. Inform. Syst. E76-D (3), 377–387.

Tallon-Baudry, C., Bertrand, O., Delpuech, C., Pernier, J., 1996. Stimulus specificity of phase-locked and non-phase-locked 40 Hz visual responses in human. J. Neurosci. 16, 4240– 4249. Tallon-Baudry, C., Bertrand, O., Peronnet, F., Pernier, J., 1998. Induced gamma-band activity during the delay of a visual short-term memory task in humans. J. Neurosci. 18, 4244– 4254. Vanhatalo, S., Palva, J.M., Holmes, M.D., Miller, J.W., Voipio, J., Kaila, K., 2004. Infraslow oscillations modulate excitability and interictal epileptic activity in the human cortex during sleep. Proc. Natl. Acad. Sci. U.S.A. 101, 5053–5057.