R

Ri

CJa

b

c

a

ARRAA

KMRTCD

C

Ers

0h

Brain Research Bulletin 97 (2013) 39– 47

Contents lists available at ScienceDirect

Brain Research Bulletin

jo ur nal homep age: www.elsev ier .com/ locate /bra inresbul l

eview

esponsive neurostimulation for the treatment of medicallyntractable epilepsy

hong Liua, Xiong-Wei Wenb, Yan Gea, Ning Chenc, Wen-Han Hua, Tan Zhangc,ian-Guo Zhangc, Fan-Gang Menga,∗

Beijing Neurosurgical Institute, Capital Medical University, Beijing, ChinaTsinghua University, Beijing, ChinaBeijing Tantan Hospital, Capital Medical University, Beijing, China

r t i c l e i n f o

rticle history:eceived 26 February 2013eceived in revised form 10 May 2013ccepted 16 May 2013vailable online 2 June 2013

eywords:edically intractable epilepsy

esponsive neurostimulationargetsurrent status

a b s t r a c t

With an annual incidence of 50/100,000 people, nearly 1% of the population suffers from epilepsy.Treatment with antiepileptic medication fails to achieve seizure remission in 20–30% of patients. Onetreatment option for refractory epilepsy patients who would not otherwise be surgical candidates iselectrical stimulation of the brain, which is a rapidly evolving and reversible adjunctive therapy. Thera-peutic stimulation can involve direct stimulation of the brain nuclei or indirect stimulation of peripheralnerves. There are three stimulation modalities that have class I evidence supporting their uses: vagusnerve stimulation (VNS), stimulation of the anterior nuclei of the thalamus (ANT), and, the most recentlydeveloped, responsive neurostimulation (RNS). While the other treatment modalities outlined deliverstimulation regardless of neuronal activity, the RNS administers stimulation only if triggered by seizureactivity. The lower doses of stimulation provided by such responsive devices can not only reduce power

evelopment consumption, but also prevent adverse reactions caused by continuous stimulation, which include thepossibility of habituation to long-term stimulation. RNS, as an investigational treatment for medicallyrefractory epilepsy, is currently under review by the FDA. Eventually systems may be developed to enableactivation by neurochemical triggers or to wirelessly transmit any information gathered. We review themechanisms, the current status, the target options, and the prospects of RNS for the treatment of medicallyintractable epilepsy.

© 2013 Elsevier Inc. All rights reserved.

ontents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 402. Stimulus mode of RNS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403. Mechanism of RNS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414. Current status of RNS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

4.1. Animal experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414.2. Clinical research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

5. Targets for RNS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 436. Prospects for RNS in the treatment of refractory epilepsy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

Conflict of interests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Abbreviations: AD, afterdischarge; ANT, anterior nuclei of the thalamus; CMT, centromCoG, electrocorticogram; EEG, electroencephalogram; HFES, high-frequency electrical sadial basis function; RNS, responsive neurostimulation; rPMC, right primary motor coeizure-like event; STN, subthalamic nucleus; SUDEP, sudden unexpected death in epilep∗ Corresponding author at: Beijing Neurosurgical Institute, Capital Medical University, N

E-mail addresses: liuchong19860810@hotmail.com (C. Liu), fgmeng@ccmu.edu.cn, me

361-9230/$ – see front matter © 2013 Elsevier Inc. All rights reserved.ttp://dx.doi.org/10.1016/j.brainresbull.2013.05.010

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

edian nucleus of the thalamus; CN, caudate nucleus; DBS, deep brain stimulation;timulation; MRF, mesencephalic reticular formation; PTZ, pentylenetetrazol; RBF,rtex; SANTE, stimulation of the anterior nucleus of thalamus for epilepsy; SLE,

sy; VNS, vagus nerve stimulation.o. 6 Tantanxili, Beijing 100050, China. Tel.: +86 1067096767; fax: +86 1067057507.ngfangang@126.com (F.-G. Meng).

4 arch Bulletin 97 (2013) 39– 47

1

a1Smppf(

tif2rahatrla

sfutT(tudpa

2

tcTetib

Mgetandtutooi

stW

half-waves, which are defined as ECoG segments between relativemaxima and minima. When the specified number of half-waves ofthe correct duration and amplitude are detected within a specified

0 C. Liu et al. / Brain Rese

. Introduction

Epilepsy is a common chronic neurological disorder that, with annnual incidence of 50 per 100,000 people, directly affects almost% of the world’s population (Sander, 2003; Brodie et al., 1997;un et al., 2008). Despite a number of treatment options [e.g. phar-acotherapy, surgery, and vagus nerve stimulation (VNS)], many

atients continue to suffer seizures. Uncontrolled epileptic attacksrofoundly impact quality of life, with major disruptions to theamilial, social, educational, and vocational activities of patientsGoldstein and Harden, 2000; Ettinger et al., 2010).

While seizure remission is achieved with antiepileptic medica-ion in between 70% and 80% of epilepsy patients, the remainder,n whom symptoms are refractory to medications, currently haveew alternative treatment options (Fridley et al., 2012; Sander,003). For these patients, one potentially curative option involvesesection of the epileptic focus when it can be clearly delineatednd safely resected (Wiebe et al., 2001). However, patients whoave seizures arising from regions of the eloquent cortex, or whichre multifocal, bilateral, or generalized, are not suitable for resec-ive epilepsy surgery. Such intractable epilepsy, which cannot beesolved with drugs or surgery, is a significant public health prob-em that necessitates the development of alternative therapeuticpproaches (Kahane and Depaulis, 2010).

One such approach involves electrical stimulation of the nervousystem, which offers a reversible, adjunctive therapeutic optionor patients with medically refractory epilepsy. Therapeutic stim-lation can either be direct or indirect, involving, for example,he stimulation of brain nuclei or peripheral nerves, respectively.hree different approaches are supported by class I evidence: VNSHandforth et al., 1998), stimulation of the anterior nuclei of thehalamus (ANT) (Fisher et al., 2010), and responsive neurostim-lation (RNS) (Morrell, 2011). Of these, RNS is the most recentlyeveloped (Wu and Sharan, 2012; Kunieda et al., 2012). In theresent article, we review the mechanisms, and current and futurepplications of RNS in the treatment of epilepsy.

. Stimulus mode of RNS

In epilepsy, because abnormal brain patterns emerge intermit-ently, the best solution would involve a closed-loop feedbackontrol system, as in RNS, that does not affect other brain functions.he closed-loop RNS device was specifically designed to recordlectrographic activity using subdural and depth electrodes in ordero detect nascent seizures and halt seizure propagation by stimulat-ng the epileptic focus, thereby terminating the embryonic seizuresefore they become clinically apparent (Fig. 1) (Berényi et al., 2012).

The implanted closed-loop NeuroPace RNS system (NeuroPace,ountain View, CA, USA) consists of a cortical strip lead, a pro-

ramming system, a pulse generator, and a depth lead (Fountast al., 2005). Responsive brain stimulation first involves implan-ation of subdural, or depth, electrodes in the target area, whichre then connected to a small device that is implanted subcuta-eously. Stimulation is delivered exclusively in response to seizureetection and relies on the programmed parameters (Fig. 2). Unlikeraditional open-loop systems, the electrodes both record and stim-late: the implanted computer continuously monitors and recordshe electrocorticogram (ECoG) at the target, and, upon detectionf an abnormal signal, the focal point is electrically stimulated inrder to disrupt the abnormal activity. Empirical evidence supportsts feasibility (Morrell, 2011; Fountas and Smith, 2007).

In the RNS system, one tool measures the length of the ECoGignal while another measures its area. The line length and areaools compare the recent average with the longer-term trend.

hen the activity within the recent window exceeds the average

Fig. 1. The flow chart of the responsive stimulation system.

trend activity by a specified percentage, detection occurs. Theline length tool is more commonly used to detect activity thatdoes not diverge from the isoelectric baseline for significant timeintervals but has a significant summed line length. On the otherhand, the area tool is more commonly used for slower rhythmicelectrographic seizure onsets that diverge from the baseline forlonger periods of time and hence have large integrated areas.The half-way tool measures the duration and the amplitude of

Fig. 2. The responsive stimulation system.

arch B

wfiaa

a1edmaIpcc

3

edffta

D1bitc

(laorlsetsmaw

bpbafsfresbrmSnicc

C. Liu et al. / Brain Rese

indow, detection also occurs. Crucially, the sensitivity and speci-city of the RNS system in the detection of actual epileptogenicctivity when all analysis tools (line length, area, and half-wave)re employed should be 100% (Fountas et al., 2005).

The delivery charge of the RNS biphasic pulse is 0.5–12 mA inmplitude, the pulse width is 40–1000 �s, and the frequency is–333 Hz (Morrell, 2011). The electrode contacts or the pulse gen-rator can be programmed as either anode or cathode. Followingelivery of a therapeutic pulse-train, a redetection algorithm deter-ines whether the epileptiform activity is still present or not. If it is,

maximum of 4 additional therapies may be delivered per episode.n addition, each therapy may consist of one or two bursts and thearameters of each therapy and/or burst may vary. The integratedharge density limit of the RNS allows up to 25 �C/cm2/phaseharge to be delivered to the patient (Fountas et al., 2005).

. Mechanism of RNS

The advantage of RNS lies in the physical combination ofpilepsy prediction and electronic stimulation, for both cortical andeep brain stimulation (DBS). The RNS acutely suppresses gammarequency (35–100 Hz) phase-locking in order to suppress epilepti-orm activity and disconnect stimulated regions from downstreamargets in epilepsy and other neuropsychiatric conditions (Sohalnd Sun, 2011).

As early as the 1990s, Durand and colleagues (Nakagawa andurand, 1991; Kayyali and Durand, 1991; Warren and Durand,998) successfully suppressed in vitro spontaneous interictal burstsy delivering RNS directly to the epileptogenic region. Their results

ndicated that the suppression mechanism is an inhibitory polariza-ion caused by the application of pulse-generated transmembraneurrents.

Afterdischarges (ADs) were used as a seizure model by LesserLesser et al., 1999) who reported in 1999 that brief 50 Hz stimu-atory bursts inhibit ADs in humans, which are terminated by thepplication of the identical stimulus for a shorter period. Continu-us rhythmic activity is more likely to terminate ADs consisting ofhythmic spiking, a stimulus at a latency of less than 4.5 s was moreikely to terminate ADs, and stimulation at the primary epilepticite was more likely to be successful (Motamedi et al., 2002). Theffect of stimulation of the primary motor cortex and supplemen-ary motor areas on AD termination during functional mapping wastudied in two patients. The latency of stimulation did not deter-ine termination success, but the numbers of channels involved

t the time of the terminating stimulus were inversely associatedith termination (Jobst et al., 2010).

Cortical RNS has been hypothesized to reduce seizure activityy inducing changes in cell membrane channels that help arrestropagation of depolarization, by altering feedback pathways, ory activating inhibitory fibers and neurotransmitters (Montgomerynd Baker, 2000; Franaszczuk et al., 2003; Sun et al., 2008). Asor DBS, responsive stimulation—particularly of mesial temporaltructures—has been shown to be beneficial in reducing seizurerequency. To test the efficacy of seizure-like event (SLE) control,adial basis function (RBF) networks that generated outputs mod-ling interictal time series were recorded from low Mg2+/high K+

olution-perfused rodent hippocampal slices and compared to RBF-ased interictal modulation, periodic biphasic-pulse modulation,andom modulation, and random repetitive modulation perfor-ance in a cognitive rhythm generator model of spontaneous SLEs.

LE moderation for the RBF interictal modulation case was sig-

ificantly improved compared to the periodic and random cases,

ndicating that biologically inspired neuromodulators may betterontrol seizures in a clinical setting (Colic et al., 2011). Thus, itan be demonstrated that high-complexity, biologically inspired

ulletin 97 (2013) 39– 47 41

responsive neuromodulation is superior to periodic forms of neu-romodulation (responsive and non-responsive), like those usedin DBS, as well as neuromodulation using random and randomrepetitive-interval stimulation.

It is also suggested that the negative feedback may be dissipat-ing energy at or near the focal area, thereby reducing excitability.Alternatively, the injected current could be altering the electro-physiological dynamics of the neurons and changing their firingpatterns (Colpan et al., 2007). Another view is that a plausiblecause of seizures is a pathology in the internal feedback action:normal internal feedback quickly regulates an abnormally highcoupling between the neural populations, whereas pathologicalinternal feedback can lead to “seizure”-like high-amplitude oscil-lations. Several external seizure control paradigms, which act toachieve the operational objective of maintaining normal levels ofsynchronous behavior, have been developed and are explored inthis paper. In particular, closed-loop “modulating” control withpredefined stimuli, and closed-loop feedback decoupling controlare considered. Of these, feedback decoupling control is the mostconsistently successful and robust seizure control strategy. The pro-posed model and remedies are consistent with a variety of recentobservations in the human and animal epileptic brain, as well aswith theories from nonlinear systems, adaptive systems, optimiza-tion, and neurophysiology. The analysis of these models has two keyconsequences: the development of a basic theory for epilepsy andother brain disorders, and the development of a robust seizure con-trol device through electrical stimulation and/or drug interventionmodalities (Chakravarthy et al., 2009).

4. Current status of RNS

The validity of RNS has been shown in recent decades throughnumerous animal experiments and clinical studies (Tables 1 and 2).On the basis of the available data, the RNS system seems to showsignificant, though modest, antiepileptic efficacy (Morrell, 2011).The RNS system is capable of real-time seizure detection and deliv-ers responsive electrical stimulation.

4.1. Animal experiments

In 1983, Psatta used feedback stimulation (5 cycles/s, 0.3 ms,1–5 V, 1–3 s) of the caudate nucleus (CN) to combat the interictalspiking activity of epileptic foci that were automatically detected bya neuroanalyzer. During 2–3 weeks training of cats, spike depres-sion followed stimulation detected at the chronic epileptogenicfocus (induced by cobalt application to the exposed neocortex). RNSwas found to be more effective than random stimulation (Psatta,1983).

Wagenaar et al. (2005) hypothesized that the persistenceof bursting is caused by a lack of input from other brainareas. Distributing stimuli across several electrodes (100–900 mV,400 �s/phase), as well as continuously fine-tuning stimulusstrength with closed-loop feedback, greatly enhanced burst con-trol. The authors concluded that externally applied electricalstimulation can substitute for natural inputs to cortical neuronalensembles in transforming burst-dominated activity to dispersedspiking, which is more reminiscent of the awake cortex in vivo. Thisnon-pharmacological method of controlling bursts will be a criticaltool for exploring the information processing capacities of neuronalensembles in vitro.

Some studies contend that more complicated control algo-

rithms for generating feedback stimulation may further improveseizure suppression. The template for a feedback signal, which isthe intracranial EEG activity at a seizure focus, is modified andthen applied directly to the seizure focus. A proportional feedback

42 C. Liu et al. / Brain Research Bulletin 97 (2013) 39– 47

Table 1Responsive stimulation for epilepsy control in animals.

Authors Year Target Animal Epilepsy model Frequency Parameter Result

Psatta 1983 CN Cat Chronic experimentalfocal epilepsy

5 cycles/s 1–5 V, 0.3 ms, 1–3 s *

Wagenaar et al. 2005 Cultured neurons Rat Dissociated neurons 10 stim/s 100–900 mV, 400 �s/phase Bursting suppress >50%Colpan et al. 2007 rPMC Rat Episodic seizures 0–50 Hz According to the different

gains of EEG in real-timeSignificant reduction inEEG amplitude

Good et al. 2009 CMT Rat Lithium-pilocarpine 130 Hz, 200 Hz 100–600 �A >50% in 33% of the ratsRajdev et al. 2011 Hippocampus Rat Kainate 5 Hz, 60 Hz, 130 Hz 150 �A, 300 �A, 5 s, 240 �s,

60 �s, 240 �s

**

Nelson et al. 2011 Somatosensorycortices

Rat Genetic absenceepilepsy rats

130 Hz, 500 Hz, 1 kHz 0.2–0.5 s, 300 �s, 500 �s,1 ms

***

Wang et al. 2012 Cortex Rat Intractable temporallobe epilepsy

1 Hz 300 ms, 0.1 mA ****

Parameters include voltage, pulse width, and amplitude.Abbreviations: MRF, mesencephalic reticular formation; rPMC, right primary motor cortex.

* Feedback CN stimulation reduced the focal spiking and the clinical seizures recorded on the EEG (p < 0.01).** Mean time for seizure to stop according to the different parameters: from 18.86 ± 16.41 s to 67.46 ± 56.56 s.

s whently l

3

swtfsirrimssIs3

fs(o(f

TC

P*

t*

*

*

d

*** Seizure durations were significantly shorter with all three stimulation strategie**** The average number of seizures in the RSG group was 15.14 ± 6.39, significa4.86 ± 6.31 of the non-stimulation group.

timulation system was designed by Colpan. Current stimulationas applied at the seizure focus by using the intracranial EEG as

he current-stimulus template. The result was that the effect ofeedback stimulation on seizures was initially assessed by mea-uring changes in the variance of the amplitude histogram of thentracranial EEG before and during stimulation, with a significanteduction found in EEG amplitude. More complicated control algo-ithms for generating feedback stimulation may provide furthermprovements in seizure suppression (Colpan et al., 2007). An auto-

ated, just-in-time stimulation, seizure control method using aeizure prediction method from nonlinear dynamics coupled withtimulation was also designed and implemented (Good et al., 2009).t showed reduction of seizure frequency and duration in five of theix rats, with significant reduction of seizure frequency (>50%) in3% of the animals.

The stimulation frequency has also been studied and it wasound that as the stimulation frequency and pulse width increases,maller pulse amplitudes are capable of eliciting an afterdischargeAD). Rajdev et al. (2011) retrospectively examined the effects

f stimulation frequency (low, medium, and high), pulse widthlow and high) and amplitude (low and high) in seizures recordedrom 23 kainic acid-treated rats. Of the stimulation parameters

able 2linical research into responsive stimulation for epilepsy.

Authors Year ILAE Target

Osorio et al. 2001 Automated seizure ANT

Kossoff et al. 2004 Complex partial seizure Cortical

Fountas et al. 2005 Complex partial seizure Different anatomicaltargets

Osorio et al. 2005 Automated seizure Epileptogeniczone/ANT

Osorio et al. 2007 Complex partial seizure Thalamic

Anderson et al. 2008 Complex partial seizure Cortical

Smith et al. 2010 Complex partial seizure Left anterior andposterior orbitofrontal

Morrell et al. 2011 Complex partial seizure Epileptogenic zone

arameters include voltage, pulse width, and amplitude.Stimulation effects can be detected and estimated with trials of this type. The only stimuhe period with the largest number of stimulated and nonstimulated seizures.*This study was not designed to test efficacy of a permanently implanted responsive cort**Seven patients (87.5%) had more than 45% decreases in their seizure frequency.***Seizures were significantly reduced in the treated group (37.9%, n = 97) compared to thifference between the treated and sham groups in adverse events.

n compared to non-stimulated (control) seizures.ower than the 27.43 ± 7.3 observed in the open-loop stimulation group and the

evaluated, the authors noted that stimulation frequency, not thecharge alone, is a key factor in the inhibition of seizure activity andthat increased frequencies and pulse widths enable smaller pulseamplitudes to elicit ADs. Thus, stimulation parameter optimizationis likely to permit the development of more efficient devices withfewer side-effects, in particular when combined with seizure pre-diction or detection algorithms in a closed-loop control application.

Nelson et al. (2011) created a closed-loop system for the auto-matic detection and control of epileptic seizures. In this preliminarystudy, a set of four EEG features was used to detect seizures andthree different electrical stimulation strategies [standard (130 Hz),very high (500 Hz) and ultra-high (1 kHz)] were delivered to termi-nate seizures. They used the mean seizure duration of epileptiformdischarges persisting beyond the end of electrical stimulation asa measure of stimulus efficacy. When compared to the dura-tion of seizures stimulated in the standard approach (7.0 s ± 10.1),both very high- and ultra-high-frequency stimulation strategieswere more effective at reducing seizure durations (1.3 ± 2.2 s and3.5 ± 6.4 s, respectively). The very high- and ultra-high-frequency

stimulation strategies may have contributed to the improvement inseizure abatement performance when compared to standard elec-trical stimulation approaches.

No. ofPatients

Frequency(Hz)

Parameter Seizurereduction

1 50–300 100–200 �s/phase,1–10 mA/phase,30 �C/cm2/phase

*

4 1–200 0.5–12 mA, 40–1000 �s **

8 1–333 0.5–12 mA, 40–1000 �s ***

8 100–500 100 �s, 5 mA 55.5%/40.8%

4 175 4.1 V, 90 �s 75.6%7 1–333 0.5–12 mA, 40–1000 �s 50%–70%1 1–333 0.5–12 mA, 40–1000 �s 60%

191 1–333 0.5–12 mA, 40–1000 �s 37.9%; ****

lation parameters that decreased seizure intensity significantly were those used in

ical neurostimulator in ambulatory patients with epilepsy.

e sham group (17.3%, n = 94; p = 0.012) during the blinded period and there was no

arch B

caiswfhpccwwe(glatnsp

4

ipqOe1lwltf2ct(lttwasiasH

qsTms1uprr12

f

C. Liu et al. / Brain Rese

A portable closed-loop brain computer interface for seizureontrol was reported in 2010 (Liang et al., 2010). It had severaldvantages, including a high seizure detection rate (92–99% dur-ng wake-sleep states), a low false detection rate (1.2–2.5%), and amall size. The seizure detection and electrical stimulation latenciesere less than 0.6 s after seizure onset. A wireless communication

eature also provided flexibility for subjects, freeing them from theassle of wires. Experimental data from freely moving rats sup-orted the functional possibility of a real-time closed-loop seizureontroller. Using this system, Liang studied closed-loop seizureontrol in epileptic rat models. The closed-loop seizure controlleras applied to three Long–Evans rats with spontaneous spike-ave discharges (non-convulsive) and three Long–Evans rats with

pileptiform activities (convulsive) induced by pentylenetetrazolPTZ) injection for evaluation. The seizure detection accuracy wasreater than 92% (up to 99%), and the average seizure detectionatency was less than 0.6 s for both spontaneous non-convulsivend PTZ-induced convulsive seizures. The average false stimula-ion rate was 3.1%. Nearly 30% of PTZ-induced convulsive seizureseeded more than two bouts of 0.5 s electrical stimulation foruppression and 90% of the non-convulsive seizures could be sup-ressed by just a single 0.5 s electrical stimulus (Liang et al., 2011).

.2. Clinical research

Clinical trials of closed-loop stimulation have been conductedn patients undergoing evaluation with intracranial electrodes asart of epilepsy surgery evaluation. Stimulations at different fre-uencies and of different targets have been tried. For example,sorio et al. (2001), in order to detect and estimate stimulationffects, used feedback stimulations (50–300 Hz, 100–200 �s/phase,–10 mA/phase, 30 �C/cm2/phase) of the ANT. The only stimu-

ation parameters that decreased seizure intensity significantlyere those used in the period with the largest number of stimu-

ated and nonstimulated seizures. The in-hospital ultra-short-termrials showed that closed-loop delivery of electrical stimulationor seizure blockage is both safe and efficacious (Osorio et al.,001). Furthermore, they studied the feasibility, safety, and effi-acy of high-frequency electrical stimulation (HFES; 100–500 Hz)riggered by automated seizure detection. There were 1491 HFESs0.2% of which triggered ADs). The mean seizure rate change in theocal closed-loop group was −55.5% (from −100% to +36.8%) andhree of four responders had a mean change of −86% (from −100%o −58.8%). In the remote closed-loop, the mean seizure rate changeas −40.8% (from −72.9% to +1.4%) and two of four responders had

mean change of −74.3% (from −75.6% to −72.9%). The mean effectize ranged from negligible in the remote closed-loop group to zeron the local closed-loop. The effects of HFES were instantaneous andlso lasted longer than the stimulation period, demonstrating theuccess of automated HFES in seizure control, and suggesting thatFES may be useful in treating pharmaco-resistant epilepsies.

The ability of the NeuroPace RNS system to reduce seizure fre-uency in 191 patients with refractory partial-onset seizures wastudied in a multicenter, double-blind, randomized, controlled trial.wo months later, with the seizure detection parameters opti-ized, patients were randomized to RNS or pure detection of

eizures without stimulation. The inclusion criteria were: age of8–70 years, partial seizures, medically refractory epilepsy (fail-re of ≥2 antiepileptic drugs), an average of three or more seizureser month, and an EEG workup showing one or two epileptogenicegions. Over a 3-month follow-up period, stimulated patientseported a 37.9% decrease in seizure frequency, which compared to

7.3% in the sham-treated group (Kossoff et al., 2004; Fountas et al.,005; Anderson et al., 2008; Smith et al., 2010; Morrell, 2011).

The safety of the implantable responsive neurostimulatoror epilepsy has been evaluated for the treatment of intractable

ulletin 97 (2013) 39– 47 43

partial-onset epilepsy in adults in clinical trials. In one trial, of the191 patients, 9 (5%) suffered intracranial hemorrhage – 6 of whichwere related to implantation – but had no permanent neurologicalsequelae. In addition, four patients suffered Sudden UnexpectedDeath in Epilepsy (SUDEP) and one committed suicide, with theRNS system enabled in all but one patient. The SUDEP rate reportedin this small population sample (11.8 per 1000 patient-years) iswithin the highest range of incidence reported for refractoryepilepsy, possibly reflecting the severity of epilepsy in the selectedpopulation (Wu et al., 2012; Morrell, 2011). However, the efficacyof this new treatment modality remains to be determined infurther multi-institutional, prospective clinical studies.

5. Targets for RNS

The key premise of RNS is the prediction of epileptic seizures,while the key to prediction is the correct choice of target. Seizurescan primarily be predicted by epileptic foci, either in cortical or deepbrain structures such as the hippocampus and its related structures(Sobayo et al., 2011; McLachlan, 2009; Morrell, 2011).

An integrated system that performs real-time electrographicanalyses and an automatic delivery of stimulation in response todetected events is required in RNS systems. Evidence from thesepreliminary studies suggests that temporal and spatial specificityare vital, and that early detection and accurate lead placement arecritical to the success of RNS. For any RNS system, the fundamen-tal question is where to place the afferent (detection) and efferent(stimulation) electrodes. The electrode number and location maybe decisive in the early detection of an impending seizure and thesubsequent application of a locally restricted stimulation (so thatthe patient remains unaware). At its most basic, the system wouldcontain a single depth-recording electrode and a processing unit;the stimulation would be delivered via the recording electrode ata critical time to cause the local brain state to revert from that ofa preictal or pro-convulsive condition to a more stable, nonicto-genic state (Peters et al., 2001; Kossoff et al., 2004; Mormann et al.,2007).

The cerebral cortex is the main RNS locus in epilepsy predic-tion because the NeuroPace RNS system uses the cortex’s electricalsignal both to predict epileptic seizures and as the stimulation tar-get. The earliest report of the use of electrical stimulation of thebrain to treat seizures in humans is from 1954 (Penfield and Jasper,1954). The researchers found that focal electrical stimulation ofthe exposed cortex could cause a flattening of the local electro-corticography (including both normal rhythms and spontaneousepileptiform discharges). Responsive cortical electrical stimulationcan effectively inhibit epileptogenic zones, reducing seizure fre-quency in patients with drug refractory epilepsy.

The hippocampus, where the Papez circuit begins and ends, isone of the targets of RNS (Shah et al., 2012). It plays an impor-tant role and is closely related to temporal lobe epilepsy (Stanslaskiet al., 2012). The Papez loop is comprised of the thalamus, hypo-thalamus, cingulate cortex, and hippocampus, and has numerouscontacts with the primary motor cortex. It is important in the reg-ulation of cerebral cortex and limbic system activity (Shah et al.,2012). Enatsu et al. (Enatsu et al., 2012) used feedback stimulationof the bilateral mesial temporal depth electrodes in the treat-ment of bilateral medial temporal lobe epilepsy, achieving a seizurefrequency reduction of 50%. Furthermore, an ANT lobectomy pre-vented the appearance of epileptic seizures.

The electrical stimulation of the anterior nucleus of thalamus

for epilepsy (SANTE) trial (Clinical Trials. Gov. NCT00101933) wasa double-blind, randomized, prospective clinical trial of DBS ofthe ANT. After a 3-month follow-up, treated patients showed amedian 40.4% decrease in seizures, compared with 14.5% in the

44 C. Liu et al. / Brain Research Bulletin 97 (2013) 39– 47

Table 3Methods of seizure prediction.

Prediction methods Authors Year Patients Seizures Sensitivity (%) Mean prediction time

Correlation dimension Lehnertz and Elger 1998 16 16 94 12 minLehnertz et al. 2001 59 95 47 19 minAschenbrenner-Scheibe et al. 2003 21 88 34 –Harrison et al. 2005b 20 960 0 –Mirowski et al. 2009 15 15 71 –Tito et al. 2009 11 126 92.31 –Rabbi et al. 2010 10 15 – –

Correlation density Martinerie et al. 1998 11 19 89 3 min

Similarity index Le Van Quyen et al. 1999 13 23 83 6 minLe Van Quyen et al. 2000 9 17 94 4 minLe Van Quyen et al. 2001a 23 26 96 7 minNavarro et al. 2002 11 41 83 8 minDe Clercq et al. 2003 12 – 0 –Winterhalder et al. 2003 21 88 42 –Navarro et al. 2005 13 129 64 >13 min

Phase synchronization Mormann et al. 2000 2 3 100 –Le Van Quyen et al. 2001b 8 – 77 Several minMormann et al. 2003b 18 32 81 4–221 sChávez et al. 2003 2 6 – ≥30 minLe Van Quyen et al. 2005 5 52 69 187 sSchelter et al. 2006 4 20 70 –Wang et al. 2011 8 22 81.82 >30 min

Lerner density Cerf and el-Ouasdad 2000 7 9 100 –

The largest Lyapunov exponent Good et al. 2010 5 – 85.9 67.6 s

Dissimilarity measure Hively et al. 2000 – 20 100 52 minProtopopescu et al. 2001 41 46 95 –Hively and Protopopescu 2003 41 46 88 35 min

Dynamical entrainment Iasemidis et al. 2001 5 58 91 49 minIasemidis et al. 2003 5 28 83 100 minIasemidis et al. 2005 2 11 82 78 minChaovalitwongse et al. 2005 10 64 69 72 minPark et al. 2011 18 20 97.5 –

Accumulated energy Litt et al. 2001 5 30 90 19 minMaiwald et al. 2004 21 88 30 –Gigola et al. 2004 4 13 92 –Esteller et al. 2005 4 42 71 85 minHarrison et al. 2005a 5 51 0 –

Simulated neuronal cells Schindler et al. 2002 7 15 100 83 minButeneers et al. 2011 23/15 452 h/982 h 96/94 –

Synchronization/correlation Mormann et al. 2003a 10 14 86 86/102 min

Sign periodogram transform Niederhauser et al. 2003 5 31 94 5–80 s

Feature selection D‘Alessandro et al. 2003 4 46 63 3 minD‘Alessandro et al. 2005 2 19 100/13 2 minNetoff et al. 2009 9 45 77.8 5 minGood et al. 2010 5 – 85.9 67.6 minAarabi and He 2012 11 49 85.1 79.9/90.2 min

Kolmogorov entropy Van Drongelen et al. 2003 5 5 60 21 min

Marginal predictability Li et al. 2003 8 24 – –Drury et al. 2003 14 44 – 30 min

Complexity/synchrony Jouny et al. 2005 2 25 0 –Raghunathan et al. 2011 a a 87.5 –

Phase clustering Kalitzin et al. 2005 3 20 – –Kerem and Geva, 2005 2005 – – 82–86 10–30 minShahidi Zandi et al. 2010 3 14 86 20.8 minShahidi Zandi et al. 2013 20 86 88.34 22.5 min

Combined method Feldwisch-Drentrup et al. 2010 8 153 27.75 –2012

cltsi

Rabbi and Fazel-Rezai, 2012

a Freiburg database.

ontrol group. The SANTE patients were then unblinded and fol-

owed for an additional time period. After a further 2 years, thereated SANTE patients had achieved a median 56% reduction ineizure frequency, with 54% of patients obtaining a ≥50% reductionn seizures. Interestingly, significantly increased adverse events of

20 56 95.8 15.8 s

depression (14.8%) and memory impairment (13.0%) were seen in

treated patients in the blinded phase (Fisher et al., 2010).

Accordingly, the ANT can be regarded as a potential target forthe treatment of refractory epilepsy with RNS. The ANT projects toboth the frontal and temporal lobes and is part of the classic circuit

arch B

oitwlt2

tntpr

6

sb2mtndfmi(dfshmLede2cM22em2Gs2pte(e2mom2svtiobi

C. Liu et al. / Brain Rese

f Papez. Because of its integral role in the limbic system, an areantimately associated with epilepsy, many groups have attemptedo stimulate the ANT in humans in an effort to suppress seizures,ith varying degrees of success (Rolston et al., 2012). However, as

ong-term clinical follow-up of the study is lacking, the long-termreatment effects of RNS of the ANT need further study (Osorio et al.,007).

However, besides the hippocampus, several other brain struc-ures are alternative targets of RNS, such as the centromedianucleus of the thalamus (CMT), subthalamic nucleus (STN), subs-antia nigra reticulata (SNr), Caudate nucleus (CN), cerebellum,osterior hypothalamus (pHyp), and caudal zona incerta (cZi). Moreesearch is needed to validate these brain regions.

. Prospects for RNS in the treatment of refractory epilepsy

On the basis of all available data, the RNS system seems tohow significant, yet modest, antiepileptic efficacy and its risk-enefit ratio should be further investigated (Ryvlin and Rheims,012). However, though RNS is a safe and effective treatment foredically intractable partial epilepsy, there is still a long way

o go, despite the great progress. Firstly, the prediction methodeeds to be improved. Real-time, accurate, automated seizureetection can be regarded as a basic platform of the technologyor RNS. Simultaneous sensing and stimulation are required to

aximize the usable neural data, minimize time delays in respond-ng, and investigate the instantaneous response to stimulationStanslaski et al., 2012). A number of methods for seizure pre-iction have been proposed, including the linear index analysisorecasting method in the early studies. The brain can be con-idered a chaotic system, so nonlinear index analysis forecastingas been more commonly used in recent studies (Table 3). Theseethods include correlation dimension (Lehnertz and Elger, 1998;

ehnertz et al., 2001; Aschenbrenner-Scheibe et al., 2003; Harrisont al., 2005b; Mirowski et al., 2009; Tito et al., 2009), correlationensity (Martinerie et al., 1998), similarity index (Le Van Quyent al., 1999, 2000, 2001a; Navarro et al., 2002; De Clercq et al.,003; Winterhalder et al., 2003; Navarro et al., 2005), phase syn-hronization (Mormann et al., 2000; Le Van Quyen et al., 2001b;ormann et al., 2003b; Chávez et al., 2003; Le Van Quyen et al.,

005; Schelter et al., 2006), Lerner density (Cerf and el-Ouasdad,000), dissimilarity measures (Hively et al., 2000; Protopopescut al., 2001; Hively and Protopopescu, 2003), dynamical entrain-ent (Iasemidis et al., 2001, 2003, 2005a,b; Chaovalitwongse et al.,

005), accumulated energy (Litt et al., 2001; Maiwald et al., 2004;igola et al., 2004; Esteller et al., 2005; Harrison et al., 2005a),imulated neuronal cells (Schindler et al., 2002; Buteneers et al.,011), synchronization/correlation (Mormann et al., 2003a), signeriodogram transform (Niederhauser et al., 2003), feature selec-ion (D‘Alessandro et al., 2003; Aarabi and He, 2012), Kolmogorovntropy (van Drongelen et al., 2003), and marginal predictabilityLi et al., 2003; Drury et al., 2003). Over the years, the methods ofpilepsy prediction continue to be verified by scholars (Lai et al.,003; Iasemidis et al., 2005a,b). In the recent studies, more andore scholars concerned with quantitative multivariate analysis

f the efficacy (Osorio et al., 2010) or the combination of variousethods (Feldwisch-Drentrup et al., 2010; Rabbi and Fazel-Rezai,

012) of epileptic prediction. Most of these methods are, however,till in the laboratory research stage, the predicted effects of thearious forecasting methods are different, and there is lack of effec-ive clinical validation (Mormann et al., 2007). Epilepsy prediction

s not just the premise of refractory epilepsy treatment, but alsone of the fundamental characteristics of RNS. Thus, RNS will onlyecome a useful clinical treatment if effective prediction of seizures

s achieved.

ulletin 97 (2013) 39– 47 45

Secondly, the parameters of electrical stimulation should bestandardized. Either continuous stimulation or activity-inducedRNS has to be applied in the clinical treatment of refractoryepilepsy. With regard to the treatment of refractory epilepsy byRNS, the intensity, frequency, and other current stimulus param-eters are not yet uniform, and researchers still need to rely onsubjective experience when determining the stimulus intensity andfrequency, according to seizure severity and patient response tostimulation. However, it is very difficult to achieve standardiza-tion and systematization of research in order to obtain coherentguidelines. The clinical application of RNS in the treatment ofrefractory epilepsy requires more long-term, multicenter, random-ized, double-blind, controlled trials.

Thirdly, as an alternative treatment for patients with drug-resistant partial epilepsy, RNS treatment, though widely recognizedto be efficacious in some patients, has some limitations. Forexample, RNS does not work for all patients. Currently, the indi-vidual efficacy of RNS cannot be predicted preoperatively and it isunknown whether such an advance will be achieved (Meng et al.,2013; Osorio et al., 2010). Nevertheless, RNS may still be one ofthe alternative methods for the treatment of medically intractableepilepsy in the future.

Conflict of interests

The authors declare no conflict of interests.

Acknowledgements

This grant was supported partly by the National Natural ScienceFoundation of China (Grant No. 81071224, 81241048), the BeijingNatural Science Foundation, China (Grant No. 7123209), and theBeijing Health System Advanced Health Technology Talent Culti-vation Plan (Grant No. 2011-3-032).

References

Aarabi, A., He, B., 2012. A rule-based seizure prediction method for focal neocorticalepilepsy. Clinical Neurophysiology 123, 1111–1122.

Anderson, W.S., Kossoff, E.H., Bergey, G.K., Jallo, G.I., 2008. Implantation of a respon-sive neurostimulator device in patients with refractory epilepsy. NeurosurgicalFocus 25, E12.

Aschenbrenner-Scheibe, R., Maiwald, T., Winterhalder, M., Voss, H.U., Timmer, J.,Schulze-Bonhage, A., 2003. How well can epileptic seizures be predicted? Anevaluation of a nonlinear method. Brain 126, 2616–2626.

Berényi, A., Belluscio, M., Mao, D., Buzsáki, G., 2012. Closed-loop control of epilepsyby transcranial electrical stimulation. Science 337, 735–737.

Brodie, M.J., Shorvon, S.D., Canger, R., Halász, P., Johannessen, S., Thompson, P., 1997.Commission on European Affairs: appropriate standards of epilepsy care acrossEurope. ILEA. Epilepsia 38, 1245–1250.

Buteneers, P., Verstraeten, D., van Mierlo, P., Wyckhuys, T., Stroobandt, D., Raedt, R.,Hallez, H., Schrauwen, B., 2011. Automatic detection of epileptic seizures on theintra-cranial electroencephalogram of rats using reservoir computing. ArtificialIntelligence in Medicine 53, 215–223.

Cerf, R., el-Ouasdad, E.H., 2000. Spectral analysis of stereo-electroencephalograms:preictal slowing in partial epilepsies. Biological Cybernetics 83, 399–405.

Chakravarthy, N., Tsakalis, K., Sabesan, S., Iasemidis, L., 2009. Homeostasis of braindynamics in epilepsy: a feedback control systems perspective of seizures. Annalsof Biomedical Engineering 37, 565–585.

Chaovalitwongse, W., Iasemidis, L.D., Pardalos, P.M., Carney, P.R., Shiau, D.S., Sackel-lares, J.C., 2005. Performance of a seizure warning algorithm based on thedynamics of intracranial EEG. Epilepsy Research 64, 93–113.

Chávez, M., Le Van Quyen, M., Navarro, V., Baulac, M., Martinerie, J., 2003. Spatio-temporal dynamics prior to neocortical seizures: amplitude versus phasecouplings. IEEE Transactions on Biomedical Engineering 50, 571–583.

Colic, S., Zalay, O.C., Bardakjian, B.L., 2011. Responsive neuromodulators based onartificial neural networks used to control seizure-like events in a computationalmodel of epilepsy. International Journal of Neural Systems 21, 367–383.

Colpan, M.E., Li, Y., Dwyer, J., Mogul, D.J., 2007. Proportional feedback stimulation

for seizure control in rats. Epilepsia 48, 1594–1603.

D‘Alessandro, M., Esteller, R., Vachtsevanos, G., Hinson, A., Echauz, J., Litt, B.,2003. Epileptic seizure prediction using hybrid feature selection over multipleintracranial EEG electrode contacts: a report of four patients. IEEE Transactionson Biomedical Engineering 50, 603–615.

4 arch B

D

D

D

E

E

E

F

F

F

F

F

F

G

G

G

G

H

H

H

H

H

I

I

I

I

J

J

K

K

K

K

6 C. Liu et al. / Brain Rese

‘Alessandro, M., Vachtsevanos, G., Esteller, R., Echauz, J., Cranstoun, S., Worrell, G.,Parish, L., Litt, B., 2005. A multi-feature and multi-channel univariate selectionprocess for seizure prediction. Clinical Neurophysiology 116, 506–516.

e Clercq, W., Lemmerling, P., Van Huffel, S., Van Paesschen, W., 2003. Anticipationof epileptic seizures from standard EEG recordings. Lancet 361, 971.

rury, I., Smith, B., Li, D., Savit, R., 2003. Seizure prediction using scalp electroen-cephalogram. Experimental Neurology 184, 9–18.

natsu, R., Alexopoulos, A., Bingaman, W., Nair, D., 2012. Complementary effect ofsurgical resection and responsive brain stimulation in the treatment of bitem-poral lobe epilepsy: a case report. Epilepsy & Behavior 24, 513–516.

steller, R., Echauz, J., D‘Alessandro, M., Worrell, G., Cranstoun, S., Vachtsevanos, G.,Litt, B., 2005. Continuous energy variation during the seizure cycle: towards anon-line accumulated energy. Clinical Neurophysiology 116, 517–526.

ttinger, A.B., Copeland, L.A., Zeber, J.E., Van Cott, A.C., Pugh, M.J., 2010. Are psy-chiatric disorders independent risk factors for new-onset epilepsy in olderindividuals? Epilepsy & Behavior 17, 70–74.

eldwisch-Drentrup, H., Schelter, B., Jachan, M., Nawrath, J., Timmer, J., Schulze-Bonhage, A., 2010. Joining the benefits: combining epileptic seizure predictionmethods. Epilepsia 51, 1598–1606.

isher, R., Salanova, V., Witt, T., Worth, R., Henry, T., Gross, R., Oommen, K., Osorio,I., Nazzaro, J., Labar, D., et al., SANTE Study Group, 2010. Electrical stimulation ofthe anterior nucleus of thalamus for treatment of refractory epilepsy. Epilepsia51, 899–908.

ountas, K.N., Smith, J.R., 2007. A novel closed-loop stimulation system in the controlof focal, medically refractory epilepsy. Acta neurochirurgica. Supplement. 97,357–362.

ountas, K.N., Smith, J.R., Murro, A.M., Politsky, J., Park, Y.D., Jenkins, P.D., 2005.Implantation of a closed-loop stimulation in the management of medicallyrefractory focal epilepsy: a technical note. Stereotactic and Functional Neuro-surgery 83, 153–158.

ranaszczuk, P.J., Kudela, P., Bergey, G.K., 2003. External excitatory stimuli can ter-minate bursting in neural network models. Epilepsy Research 53, 65–80.

ridley, J., Thomas, J.G., Navarro, J.C., Yoshor, D., 2012. Brain stimulation for thetreatment of epilepsy. Neurosurgical Focus 32, E13.

igola, S., Ortiz, F., D‘Attellis, C.E., Silva, W., Kochen, S., 2004. Prediction of epilepticseizures using accumulated energy in a multiresolution framework. Journal ofNeuroscience Methods 138, 107–111.

oldstein, M.A., Harden, C.L., 2000. Epilepsy and anxiety. Epilepsy & Behavior 1,228–234.

ood, L.B., Sabesan, S., Marsh, S.T., Tsakalis, K., Treiman, D., Iasemidis, L., 2009. Con-trol of synchronization of brain dynamics leads to control of epileptic seizuresin rodents. International Journal of Neural Systems 19, 173–196.

ood, L.B., Sabesan, S., Marsh, S.T., Tsakalis, K., Treiman, D., Iasemidis, L., 2010. Non-linear dynamics of seizure prediction in a rodent model of epilepsy. NonlinearDynamics, Psychology, and Life Sciences 14, 411–434.

andforth, A., DeGiorgio, C.M., Schachter, S.C., Uthman, B.M., Naritoku, D.K., Tecoma,E.S., Henry, T.R., Collins, S.D., Vaughn, B.V., Gilmartin, R.C., et al., 1998. Vagusnerve stimulation therapy for partial-onset seizures: a randomized active-control trial. Neurology 51, 48–55.

arrison, M.A., Frei, M.G., Osorio, I., 2005a. Accumulated energy revisited. ClinicalNeurophysiology 116, 527–531.

arrison, M.A., Osorio, I., Frei, M.G., Asuri, S., Lai, Y.C., 2005b. Correlation dimensionand integral do not predict epileptic seizures. Chaos 15, 33106.

ively, L.M., Protopopescu, V.A., 2003. Channel-consistent forewarning of epilep-tic events from scalp EEG. IEEE Transactions on Biomedical Engineering 50,584–593.

ively, L.M., Protopopescu, V.A., Gailey, P.C., 2000. Timely detection of dynamicalchange in scalp EEG signals. Chaos 10, 864–875.

asemidis, L.D., Pardalos, P., Sackellares, J.C., Shiau, D.S., 2001. Quadratic binary pro-gramming and dynamical system approach to determine the predictability ofepileptic seizures. Journal of Combinatorial Optimization 5, 9–26.

asemidis, L.D., Shiau, D.S., Chaovalitwongse, W., Sackellares, J.C., Pardalos, P.M.,Principe, J.C., Carney, P.R., Prasad, A., Veeramani, B., Tsakalis, K., 2003. Adaptiveepileptic seizure prediction system. IEEE Transactions on Biomedical Engineer-ing 50, 616–627.

asemidis, L.D., Shiau, D.S., Pardalos, P.M., Chaovalitwongse, W., Narayanan,K., Prasad, A., Tsakalis, K., Carney, P.R., Sackellares, J.C., 2005a. Long-termprospective on-line real-time seizure prediction. Clinical Neurophysiology 116,532–544.

asemidis, L.D., Tsakalis, K., Sackellares, J.C., Pardalos, P.M., 2005b. Comment on“Inability of Lyapunov exponents to predict epileptic seizures”. Physical ReviewLetters 94, 019801, author reply 019802.

obst, B.C., Darcey, T.M., Thadani, V.M., Roberts, D.W., 2010. Brain stimulation for thetreatment of epilepsy. Epilepsia 51, 88–92.

ouny, C.C., Franaszczuk, P.J., Bergey, G.K., 2005. Signal complexity and synchronyof epileptic seizures: is there an identifiable preictal period? Clinical Neuro-physiology 116, 552–558.

ahane, P., Depaulis, A., 2010. Deep brain stimulation in epilepsy: what is next?Current Opinion in Neurology 23, 177–182.

alitzin, S., Velis, D., Suffczynski, P., Parra, J., da Silva, F.L., 2005. Electrical brain-stimulation paradigm for estimating the seizure onset site and the time to ictal

transition in temporal lobe epilepsy. Clinical Neurophysiology 116, 718–728.

ayyali, H., Durand, D., 1991. Effects of applied currents on epileptiform bursts invitro. Experimental Neurology 113, 249–254.

erem, D.H., Geva, A.B., 2005. Forecasting epilepsy from the heart rate signal. Medicaland Biological Engineering and Computing 43, 230–239.

ulletin 97 (2013) 39– 47

Kossoff, E.H., Ritzl, E.K., Politsky, J.M., Murro, A.M., Smith, J.R., Duckrow, R.B., Spencer,D.D., Bergey, G.K., 2004. Effect of an external responsive neurostimulator onseizures and electrographic discharges during subdural electrode monitoring.Epilepsia 45, 1560–1567.

Kunieda, T., Kikuchi, T., Miyamoto, S., 2012. Epilepsy surgery: surgical aspects. Cur-rent Opinion in Anaesthesiology 25, 533–539.

Lai, Y.C., Harrison, M.A., Frei, M.G., Osorio, I., 2003. Inability of Lyapunov exponentsto predict epileptic seizures. Physical Review Letters 91, 068102.

Le Van Quyen, M., Adam, C., Martinerie, J., Baulac, M., Clemenceau, S., Varela, F., 2000.Spatio-temporal characterization of non-linear changes in intracranial activitiesprior to human temporal lobe seizures. European Journal of Neuroscience 12,2124–2134.

Le Van Quyen, M., Martinerie, J., Baulac, M., Varela, F., 1999. Anticipating epilepticseizure in real time by a nonlinear analysis of similarity between EEG recordings.Neuroreport 10, 2149–2155.

Le Van Quyen, M., Martinerie, J., Navarro, V., Baulac And, M., Varela, F.J., 2001b.Characterizing neurodynamic changes before seizures. Journal of Clinical Neu-rophysiology 18, 191–208.

Le Van Quyen, M., Martinerie, J., Navarro, V., Boon, P., D‘Have, M., Adam, C., Renault,B., Varela, F., Baulac, M., 2001a. Anticipation of epileptic seizures from standardEEG recordings. Lancet 357, 183–188.

Le Van Quyen, M., Soss, J., Navarro, V., Robertson, R., Chavez, M., Baulac, M., Mar-tinerie, J., 2005. Preictal state identification by synchronization changes inlong-term intracranial EEG recordings. Clinical Neurophysiology 116, 559–568.

Lehnertz, K., Elger, C.E., 1998. Can epileptic seizures be predicted? Evidence fromnonlinear time series analysis of brain electrical activity. Physical Review Letters80, 5019–5023.

Lehnertz, K., Andrzejak, R.G., Arnhold, J., Kreuz, T., Mormann, F., Rieke, C., 2001.Nonlinear EEG analysis in epilepsy: its possible use for interictal focus localiza-tion, seizure anticipation, and prevention. Journal of Clinical Neurophysiology18, 209–222.

Lesser, R.P., Kim, S.H., Beyderman, L., Miglioretti, D.L., Webber, W.R., Bare, M., Cysyk,B., Krauss, G., Gordon, B., 1999. Brief bursts of pulse stimulation terminate after-discharges caused by cortical stimulation. Neurology 53, 2073–2081.

Li, D., Zhou, W., Drury, I., Savit, R., 2003. Non-linear, non-invasive method for seizureanticipation in focal epilepsy. Mathematical Biosciences 186, 63–77.

Liang, S.F., Liao, Y.C., Shaw, F.Z., Chang, D.W., Young, C.P., Chiueh, H., 2011. Closed-loop seizure control on epileptic rat models. Journal of Neural Engineering 8,045001.

Liang, S.F., Shaw, F.Z., Young, C.P., Chang, D.W., Liao, Y.C., 2010. A closed-loopbrain computer interface for real-time seizure detection and control. Confer-ence Proceedings: Annual International Conference of the IEEE Engineering inMedicine and Biology Society 2010, 4950–4953.

Litt, B., Esteller, R., Echauz, J., D‘Alessandro, M., Shor, R., Henry, T., Pennell, P.,Epstein, C., Bakay, R., Dichter, M., Vachtsevanos, G., 2001. Epileptic seizures maybegin hours in advance of clinical onset: a report of five patients. Neuron 30,51–64.

Maiwald, T., Winterhalder, M., Aschenbrenner-Scheibe, R., Voss, H.U., Schulze-Bonhage, A., Timmer, J., 2004. Comparison of three nonlinear seizure predictionmethods by means of the seizure prediction characteristic. Physica D 194,357–368.

Martinerie, J., Adam, C., Le Van Quyen, M., Baulac, M., Clemenceau, S., Renault, B.,Varela, F.J., 1998. Epileptic seizures can be anticipated by non-linear analysis.Nature Medicine 4, 1173–1176.

McLachlan, R.S., 2009. A brief review of the anatomy and physiology of the limbicsystem. Canadian Journal of Neurological Sciences 36 (Suppl. 2), S84–S87.

Meng, F.G., Kao, C.C., Zhang, H., Chen, N., Ge, Y., Liu, C., Zhang, J.G., 2013. Usingelectroacupuncture at acupoints to predict the efficacy of hippocampal high-frequency electrical stimulation in pharmacoresistant temporal lobe epilepsypatients. Medical Hypotheses 80, 244–246.

Mirowski, P., Madhavan, D., Lecun, Y., Kuzniecky, R., 2009. Classification of patternsof EEG synchronization for seizure prediction. Clinical Neurophysiology 120,1927–1940.

Montgomery Jr., E.B., Baker, K.B., 2000. Mechanisms of deep brain stimulation andfuture technical developments. Neurological Research 22, 259–266.

Mormann, F., Andrzejak, R.G., Elger, C.E., Lehnertz, K., 2007. Seizure prediction: thelong and winding road. Brain 130, 314–333.

Mormann, F., Andrzejak, R.G., Kreuz, T., Rieke, C., David, P., Elger, C.E., Lehnertz,K., 2003a. Automated detection of a pre-seizure state based on a decrease insynchronization in intracranial EEG recordings from epilepsy patients. PhysicalReview E: Statistical, Nonlinear, and Soft Matter Physics 67, 021912.

Mormann, F., Kreuz, T., Andrzejak, R.G., David, P., Lehnertz, K., Elger, C.E., 2003b.Epileptic seizures are preceded by a decrease in synchronization. EpilepsyResearch 53, 173–185.

Mormann, F., Lehnertz, K., David, P., Elger, C.E., 2000. Mean phase coherence as ameasure for phase synchronization and its application to the EEG of epilepsypatients. Physica D 144, 358–369.

Morrell, M.J., RNS System in Epilepsy Study Group, 2011. Responsive cortical stimu-lation for the treatment of medically intractable partial epilepsy. Neurology 77,1295–1304.

Motamedi, G.K., Lesser, R.P., Miglioretti, D.L., Mizuno-Matsumoto, Y., Gordon, B.,

Webber, W.R., Jackson, D.C., Sepkuty, J.P., Crone, N.E., 2002. Optimizing param-eters for terminating cortical afterdischarges with pulse stimulation. Epilepsia43, 836–846.

Nakagawa, M., Durand, D., 1991. Suppression of spontaneous epileptiform activitywith applied currents. Brain Research 567, 241–247.

arch B

N

N

N

N

N

O

O

O

O

P

P

P

P

R

R

R

R

R

R

S

S

C. Liu et al. / Brain Rese

avarro, V., Martinerie, J., Le Van Quyen, M., Baulac, M., Dubeau, F., Gotman, J.,2005. Seizure anticipation: do mathematical measures correlate with video-EEGevaluation? Epilepsia 46, 385–396.

avarro, V., Martinerie, J., Le Van Quyen, M., Clemenceau, S., Adam, C., Baulac, M.,Varela, F., 2002. Seizure anticipation in human neocortical partial epilepsy. Brain125, 640–655.

elson, T.S., Suhr, C.L., Freestone, D.R., Lai, A., Halliday, A.J., McLean, K.J., Burkitt,A.N., Cook, M.J., 2011. Closed-loop seizure control with very high frequencyelectrical stimulation at seizure onset in the GAERS model of absence epilepsy.International Journal of Neural Systems 21, 163–173.

etoff, T., Park, Y., Parhi, K., 2009. Seizure prediction using cost-sensitive supportvector machine. Conference Proceedings: Annual International Conference ofthe IEEE Engineering in Medicine and Biology Society 2009, 3322–3325.

iederhauser, J.J., Esteller, R., Echauz, J., Vachtsevanos, G., Litt, B., 2003. Detectionof seizure precursors from depth-EEG using a sign periodogram transform. IEEETransactions on Biomedical Engineering 50, 449–458.

sorio, I., Frei, M.G., Manly, B.F., Sunderam, S., Bhavaraju, N.C., Wilkinson, S.B.,2001. An introduction to contingent (closed-loop) brain electrical stimulationfor seizure blockage, to ultra-short-term clinical trials, and to multidimensionalstatistical analysis of therapeutic efficacy. Journal of Clinical Neurophysiology18, 533–544.

sorio, I., Frei, M.G., Sunderam, S., Giftakis, J., Bhavaraju, N.C., Schaffner, S.F.,Wilkinson, S.B., 2005. Automated seizure abatement in humans using electricalstimulation. Annals of Neurology 57, 258–268.

sorio, I., Manly, B., Sunderam, S., 2010. Toward a quantitative multivariateanalysis of the efficacy of antiseizure therapies. Epilepsy & Behavior 18,335–343.

sorio, I., Overman, J., Giftakis, J., Wilkinson, S.B., 2007. High frequency thalamicstimulation for inoperable mesial temporal epilepsy. Epilepsia 48, 1561–1571.

enfield, W., Jasper, H., 1954. Electrocorticography. In: Epilepsy and the FunctionalAnatomy of the Human Brain. Little, Brown, Boston, pp. 692–738.

eters, T.E., Bhavaraju, N.C., Frei, M.G., Osorio, I., 2001. Network system for auto-mated seizure detection and contingent delivery of therapy. Journal of ClinicalNeurophysiology 18, 545–549.

rotopopescu, V.A., Hively, L.M., Gailey, P.C., 2001. Epileptic seizure forewarningfrom scalp EEG. Journal of Clinical Neurophysiology 18, 223–245.

satta, D.M., 1983. Control of chronic experimental focal epilepsy by feedback cau-datum stimulations. Epilepsia 24, 444–454.

abbi, A.F., Aarabi, A., Fazel-Rezai, R., 2010. Fuzzy rule-based seizure predic-tion based on correlation dimension changes in intracranial EEG. ConferenceProceedings: Annual International Conference of the IEEE Engineering inMedicine and Biology Society 2010, 3301–3304.

abbi, A.F., Fazel-Rezai, R., 2012. A fuzzy logic system for seizure onset detection inintracranial EEG. Computational Intelligence and Neuroscience 2012, 705140,http://dx.doi.org/10.1155/2012/705140.

aghunathan, S., Jaitli, A., Irazoqui, P.P., 2011. Multistage seizure detection tech-niques optimized for low-power hardware platforms. Epilepsy & Behavior 22(Suppl. 1), S61–S68.

ajdev, P., Ward, M., Irazoqui, P., 2011. Effect of stimulus parameters in the treatmentof seizures by electrical stimulation in the kainate animal model. InternationalJournal of Neural Systems 21, 151–162.

olston, J.D., Englot, D.J., Wang, D.D., Shih, T., Chang, E.F., 2012. Comparison ofseizure control outcomes and the safety of vagus nerve, thalamic deep brain,and responsive neurostimulation: evidence from randomized controlled trials.Neurosurgical Focus 32, E14.

yvlin, P., Rheims, S., 2012. Epilepsy: new epidemiological and therapeutic perspec-tives. Lancet Neurology 11, 5–7.

ander, J.W., 2003. The epidemiology of epilepsy revisited. Current Opinion in Neu-rology 16, 165–170.

chelter, B., Winterhalder, M., Maiwald, T., Brandt, A., Schad, A., Schulze-Bonhage,A., Timmer, J., 2006. Testing statistical significance of multivariate time seriesanalysis techniques for epileptic seizure prediction. Chaos 16, 013108.

ulletin 97 (2013) 39– 47 47

Schindler, K., Wiest, R., Kollar, M., Donati, F., 2002. EEG analysis with simulated neu-ronal cell models helps to detect pre-seizure changes. Clinical Neurophysiology113, 604–614.

Shah, A., Jhawar, S.S., Goel, A., 2012. Analysis of the anatomy of the Papez circuitand adjoining limbic system by fiber dissection techniques. Journal of ClinicalNeuroscience 19, 289–298.

Shahidi Zandi, A., Tafreshi, R., Javidan, M., Dumont, G.A., 2010. Predicting temporallobe epileptic seizures based on zero-crossing interval analysis in scalp EEG. Con-ference Proceedings: Annual International Conference of the IEEE Engineeringin Medicine and Biology Society 2010, 5537–5540.

Shahidi Zandi, A., Tafreshi, R., Javidan, M., Dumont, G.A., 2013. Predicting epilepticseizures in scalp EEG based on a variational bayesian gaussian mixture modelof zero-crossing intervals. IEEE Transactions on Bio-Medical Engineering 60,1401–1413.

Smith, J.R., Fountas, K.N., Murro, A.M., Park, Y.D., Jenkins, P.D., Morrell, M.,Esteller, R., Greene, D., 2010. Closed-loop stimulation in the control offocal epilepsy of insular origin. Stereotactic and Functional Neurosurgery 88,281–287.

Sobayo, T., Fine, A.S., Mogul, D.J., 2011. A study of multi-site brain dynamics duringlimbic seizures. Conference Proceedings: Annual International Conference of theIEEE Engineering in Medicine and Biology Society 2011, 7557–7559.

Sohal, V.S., Sun, F.T., 2011. Responsive neurostimulation suppresses synchronizedcortical rhythms in patients with epilepsy. Neurosurgery Clinics of North Amer-ica 22, 481–488.

Stanslaski, S., Afshar, P., Cong, P., Giftakis, J., Stypulkowski, P., Carlson, D., Linde,D., Ullestad, D., Avestruz, A.T., Denison, T., 2012. Design and validation of afully implantable, chronic, closed-loop neuromodulation device with concurrentsensing and stimulation. IEEE Transactions on Neural Systems and RehabilitationEngineering 20, 410–421.

Sun, F.T., Morrell, M.J., Wharen Jr., R.E., 2008. Responsive cortical stimulation for thetreatment of epilepsy. Neurotherapeutics 5, 68–74.

Tito, M., Cabrerizo, M., Ayala, M., Barreto, A., Miller, I., Jayakar, P., Adjouadi, M.,2009. Classification of electroencephalographic seizure recordings into ictal andinterictal files using correlation sum. Computers in Biology and Medicine 39,604–614.

van Drongelen, W., Nayak, S., Frim, D.M., Kohrman, M.H., Towle, V.L., Lee,H.C., McGee, A.B., Chico, M.S., Hecox, K.E., 2003. Seizure anticipation inpediatric epilepsy: use of Kolmogorov entropy. Pediatric Neurology 29,207–213.

Wagenaar, D.A., Madhavan, R., Pine, J., Potter, S.M., 2005. Controlling bursting incortical cultures with closed-loop multi-electrode stimulation. Journal of Neu-roscience 25, 680–688.

Wang, L., Wang, C., Fu, F., Yu, X., Guo, H., Xu, C., Jing, X., Zhang, H., Dong, X., 2011.Temporal lobe seizure prediction based on a complex Gaussian wavelet. ClinicalNeurophysiology 122, 656–663.

Warren, R.J., Durand, D.M., 1998. Effects of applied currents on spontaneous epilep-tiform activity induced by low calcium in the rat hippocampus. Brain Research806, 186–195.

Wiebe, S., Blume, W.T., Girvin, J.P., Eliasziw, M., Effectiveness and Efficiency ofSurgery for Temporal Lobe Epilepsy Study Group, 2001. A randomized, con-trolled trial of surgery for temporal-lobe epilepsy. New England Journal ofMedicine 345, 311–318.

Winterhalder, M., Maiwald, T., Voss, H.U., Aschenbrenner-Scheibe, R., Timmer, J.,Schulze-Bonhage, A., 2003. The seizure prediction characteristic: a generalframework to assess and compare seizure prediction methods. Epilepsy &Behavior 4, 318–325.

Wu, C., Sharan, A.D., 2012. Neurostimulation for the treatment of epilepsy: a review

of current surgical interventions. Neuromodulation 16, 10–24.

Wu, C., Evans, J.J., Skidmore, C., Sperling, M.R., Sharan, A.D., 2012. Impedancevariations over time for a closed-loop neurostimulation device: earlyexperience with chronically implanted electrodes. Neuromodulation 16,46–50.