International Journal of Psychophysiology 91 (2014) 10–15

Contents lists available at ScienceDirect

International Journal of Psychophysiology

j ourna l homepage: www.e lsev ie r .com/ locate / i jpsycho

Mobile EEG and its potential to promote the theory and application ofimagery-based motor rehabilitation

Cornelia Kranczioch a,b, Catharina Zich a, Irina Schierholz a, Annette Sterr c,⁎a Neuropsychology Lab, Department of Psychology, European Medical School, Carl von Ossietzky University of Oldenburg, Oldenburg, Germanyb Neurosensory Science Research Group, Carl von Ossietzky University of Oldenburg, Oldenburg, Germanyc Brain and Behavior Research Group, School of Psychology, University of Surrey, Guildford, UK

⁎ Corresponding author at: Lewis Carroll Building, SchSurrey, Guildford GU2 7XH, UK. Tel.: +44 1483 682883; f

E-mail address: a.sterr@surrey.ac.uk (A. Sterr).

0167-8760/$ – see front matter © 2013 Elsevier B.V. All rihttp://dx.doi.org/10.1016/j.ijpsycho.2013.10.004

a b s t r a c t

a r t i c l e i n f o

Article history:Received 23 August 2013Received in revised form 8 October 2013Accepted 10 October 2013Available online 18 October 2013

Keywords:Mobile EEGWireless EEGMIMotor imageryNeurofeedbackNeurological rehabilitationBCIBrain computer interface

Studying the brain in its natural state remains a major challenge for neuroscience. Solving this challenge wouldnot only enable the refinement of cognitive theory, but also provide a better understanding of cognitive functionin the type of complex and unpredictable situations that constitute daily life, and which are often disturbed inclinical populations. With mobile EEG, researchers now have access to a tool that can help address these issues.In this paper we present an overview of technical advancements in mobile EEG systems and associated analysistools, and explore the benefits of this new technology. Using the example of motor imagery (MI) wewill examinethe translational potential ofMI-based neurofeedback training for neurological rehabilitation and applied research.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

Understanding human behavior is one of the big challenges forhumankind. In the last 20 years neuroscience has emerged as a keyarea of research and there is a recognition that understanding brainfunction in general, and brain–behavior relationships in particular, isvital to advance solutions for major public health issues such as mentalhealth, dementia, obesity, or impairments remaining after sufferingfrom stroke or traumatic brain injury. Immense improvements in theavailability of neuroimaging methodologies together with high-profileinitiatives, such as the decade of the brain, have brought a wealth ofnew insights into brain function and are already leading to new formsof treatment. However, a major challenge still is to understand thebrain in its natural state. This would not only enable the refinementof cognitive theory but also to get a true understanding of cognitivefunction in the type of complex and unpredictable situations thatconstitute daily life. For example, how does our brain enable us tofunction in a highly complex situation such as navigating through agrocery shop while selecting products from a vast range of goods?How are these processes influenced by internal physiological statessuch as hunger or low mood? How does our brain help us to prioritizesome actions and inhibit others? What are the brain correlates of

ool of Psychology, University ofax: +44 1483 682914.

ghts reserved.

impaired, challenging or maladaptive behavior expressed in typical lifesituations? These are of course hugely demanding questions, whichcannot be easily answered. Yet with the mobile electroencephalogram(EEG), researchers now have a tool to explore these questions. Incontrast to all other techniques presently available, mobile EEG trulyallows us to take neuroscience into the field and study everyday brainfunction.

In this paper we will examine the benefits of mobile EEG andthe challenges it has to meet to provide a fully fledged research toolin cognitive and clinical neuroscience, as well as a tool for clinicalinterventions and BCIs. We will exemplarily show how the technicalchallenges involved in mobile EEG have been addressed by recentadvancements in the field. The focus will then be shifted to yet anotheropportunity associated with mobile EEG, which is the support of braincomputer interface (BCI) based treatment delivery in the homeenvironment. This will be done through the example of motor imagery(MI).

2. Why do we need mobile EEG?

EEG studies are typically conducted in a laboratory, and manyarguments can be found in favor of this practice. For example, theenvironment is controlled and recording conditions are kept consistentacross subjects. Laboratories are often electrically shielded and noiseattenuated to reduce factors that may negatively affect data qualitysuch as line noise.Moreover, the laboratory set-up allows for full control

11C. Kranczioch et al. / International Journal of Psychophysiology 91 (2014) 10–15

of the amount and type of information a participant is exposed to at anytime of the experiment. In addition, participants are typically askedto keep their heads still and relax their jaws and necks so as to avoidcontamination of the EEG bymuscle artifacts. Evenmore serious sourcesof artifacts are eye movements and blinks. To avoid those, participantsare normally asked to fix their eyes on a particular point on the screenand, in many cases, also to blink only during specified periods in thetrial (also called blink holidays). All these characteristics are designedto instigate methodological rigor. However, the methodological strengthof the laboratory-based approach comes at the expense of ecologicalvalidity.

This drawback has been recognized in the research community, andlet for instance to the development of what has become known asthe mobile brain/body imaging (MoBI) approach. The MoBI approachis motivated by the ambition to understand brain activity supportingembodied human cognition, i.e. cognition linked to our own motorbehavior and that of other individuals (e.g. Makeig et al., 2009, fora recent review see Gramann et al., 2011). In MoBI, brain and bodydynamics are recorded simultaneously by combining high-densityscalp EEG and motion capture. The EEG setup works with standardamplifiers, which are fixed somewhere above the participants' head.The participant walks on a treadmill while solving cognitive tasks.Research has shown that it is possible to record good-qualitycognitive ERPs with the original MoBI (e.g. Gramann et al., 2010)and with variations of the original setup (De Sanctis et al., 2012).However, while certainly being a large step forward, due to itsreliance on traditional EEG hardware themobility of MoBI is still stronglylimited, and it therefore cannot solve the problem of ecological validitysatisfactorily.

Without doubt, laboratory studies on the neural correlates of humanbehavior have revealed many hugely important insights. However, thelaboratory setting is far removed from natural behavior, and does notadequately reflect the complexity of information processing requiredfor a person to function ‘normally’ in everyday life. For instance, tasksare typically investigated in isolation, and without the distractors apersonwould have to dealwith in the realworld, such as environmentalbackground noise. In addition, cognition and behavior are studiedwhile seated or at least when being more or less ‘stationary’, ratherthan moving around. Laboratory studies therefore strongly limit thede facto simultaneity of cognitive processes and, and critically, theneed to perform actions/having to respond with complex actions. Dataacquired in laboratory settings therefore limit our understanding ofhow the brain controls cognitive processes and behavior to situationsthat are not representative of the complex information processingassociated with natural behavior. The latter not only leaves a gapin knowledge, but also hinders the development of cognitive theoryreflecting the association of brain function and behavior morerealistically. Thus, unless we are able to move outside the laboratoryand study the brain during natural behavior, theory development isbound to be ‘circular’ in the sense that theories derived from laboratoryresearch are likely be confirmed when tested in yet another laboratorystudy.

The current advancements in the ability to record and make useof EEG signals outside the laboratory are technically fascinating, andthey will certainly be followed with great interest by those inspiredby the idea that computers can be controlled through thought. Butthe relevance of the technical advancements in mobile EEG goes farbeyond technical fascination as it allows for the first time to study howthe brain processes makes use and responds to complex informationand situations, andhence toderive anoninvasiveneuralmarker of naturalbehavior in humans. Moreover, mobile EEG will also greatly extendthe range of EEG-based therapeutic applications, and the availability ofhigh-quality BCIs, as it allows the provision of these methods outsidespecialized clinical or laboratory settings. Because of the comparativelylow costs of EEG systems and the relative ease of application, such BCIscould also be used to establish home-based therapeutic interventions

that require regular training over a long period of time but that,to a certain degree, can be run by patients themselves. We will returnto this aspect of mobile EEG later in this paper. But before doing so wewill present a short overview of the most prominent systems presentlyavailable on the market and detailed discussion of current advancementsin the use of mobile EEG in cognitive neuroscience.

3. Mobile EEG systems and current advancements

Mobile EEG systems have been available for a number of yearsnow. But which requirements shall a true mobile EEG system fulfill?Obviously mobile EEG systems should allow natural body movements,which imply the use of non-stationary EEG systems that are ideallyfully head-mounted. Systems therefore should be small, lightweightand transmit data wirelessly. Moreover a head-mounted cap-amplifierdesign ensures minimal isolated movements of individual electrodes,cables or the amplifier, whichminimizes disturbances of EEG by electro-magnetic interference and therewith dramatically improve EEG signalquality. Furthermore mobile EEG systems should first consist of asufficient number of electrodes to enable spatial filter based artifactattenuation (Makeig et al., 2009) and allow flexible placement ofthese electrodes so that the same EEG system can be applied to differentresearch questions. One possible way to increase the applicability ofmobile EEG systems even more, which is especially important for EEGbased BCIs, is the use of dry electrodes which require no conductivegel. However, since hardly any study has shown a comparable signalquality between dry and wet electrodes, especially not during grossbody movements as walking, dry electrodes are not as common as onemight expect.

Over the years, the devices develop rapidly, which not only meansthat they are becoming easier and faster to apply, but also that theybecome less costly and thus increasingly accessible. Most of the systemscurrently available are commercial systems designed for the gamingor advertising industry, or marketing research. Only a few devices areobtainable that were specifically designed for neuroscience research.

One of the first devices able to record brain activity in a mobilewireless mode was the MindWave (Neurosky Inc., San Jose, CA, www.neurosky.com). The device consists of a single EEG sensor comprisinga dry electrode, which makes the MindSet easy and fast to apply. Thedownside is that the single sensor can only be placed on the forehead,making the system very inflexible and limited in its application. Powerfor the device is supplied by an AAA battery, which, according to themanufacturer, allows a nonstop 10-hour recording. The system providesthe raw EEG data and information about the frequency content of theEEG. The additionally available software eSense claims to visualize currentlevels of attention and meditation, which are supposed to be usable forcontrolling devices (www.neurosky.com).

Another low-cost pocket-sized device has been developed by AvatarEEG Solutions Inc (Avatar EEG solutions Inc., Calgary, Canada, www.avatareeg.com). The system comes without electrodes but featureseight channels, which can be configured in several different bipolar ormonopolar montages, and which therefore allow to record EEG as wellas electrooculogram(EOG), (electromyogram) EMGor electrocardiogram(ECG). The system supports various electrode types such as surfaceelectrodes, intracranial and subdural electrodes. It is light and small(76 × 53 × 38 mm; 60 g). The power is provided by two rechargeableAA lithium batteries, which should suffice for 24 hours of continuousrecordings. The collected data can either be viewed in real-time on asmart phone over a maximal distance of 30m, or they can be written toa removable onboard Micro SD card (8GB–32GB) for subsequent offlineanalysis. The developers highlight the system's low energy consumptionand variable electrode configurations. They see applications for theirsystem primarily in sleep research, but also in ambulatory neurosciencestudies.

The system Enobio (Neuroelectrics, Barcelona, Spain, http://neuroelectrics.com/enobio) offers the possibility to record between

12 C. Kranczioch et al. / International Journal of Psychophysiology 91 (2014) 10–15

eight and 20 EEG channels. Like the other systems it is small(60×85×20mm) and low inweight (65g),whichmakes it comfortableto wear. The Enobio system has been developed with a focus onBCI research. Beyond that it is also advertised to be of value for a varietyof other applications such as basic research, medical applicationdevelopment, neuromodulation, and biometry. The system offers amaximal recording duration of about 16 hours before rechargingis required. Data are transferred wirelessly to a laptop, where the rawdata can bemonitored in real-time. The software accompanying Enobioalso includes functions such as real-time spectrograms, filtering,streaming and feature extraction. In addition, the data can be storedon an onboard Micro SD card interface that can save more than24 hours of recorded data. The system can use different kindsof electrodes, which are mounted on an EEG cap according to theinternational 10/20 system. The sampling rate of the Enobio deviceis 500 Hz, it has a resolution of 24 bits. The resulting bandwidth of0–250 Hz captures the EEG in all frequency bands relevant to humancognition.

The company Emotiv (Emotiv, Hong Kong, http://www.emotiv.com) has so far developed two different commercial systems: TheEPOC-headset and the EEG-headset, both of which are multi-channel(14 saline sensors and 2 reference electrodes)wireless-enabled systems(128Hz sampling rate; 0.16–45Hz band-pass). The devices run with alithium battery and can continuously be used for up to 12hourswithoutcharging. The data recorded can be visualized in real-time. The systemscan be linked to a computer by awireless Bluetooth connection. The twodevices are nearly identical. As an additional feature the EEG headsetoffers direct access to raw EEG data that can be viewed in real-timewith the provided specialized software TestBench. Emotiv advertisestheir systems as devises enabling brain-activity based access toapplications, e.g. for playing computer games. Concerning the use ofthe Emotiv system in research, a recent study (Badcock et al., 2013)compared it to the widely used conventional EEG system Neuroscanin an auditory paradigm. The results of this study showed that theEmotiv represents a solid alternative for measuring brain potentialsthat are relatively reliable, like late auditory evoked potentials, butnot when focusing on less reliable potentials such as the mismatchnegativity.

As this short and selective overview illustrates, the progress in thefield of mobile EEG is considerable. Yet many issues remain to be solved,among themdata quality, user friendliness, and robustness. An increasingnumber of companies are dedicated to achieve improvements herein.One of those is mBrainTrain (Belgrade, Serbia, www.mbraintrain.com),a company specializing in the advancement of fully mobile and wirelessEEG systems. The company promises a mobile and wireless EEG systemrunning on notebooks and smart phones using a small 24 channelamplifier with similar characteristics to a stationary laboratory amplifier(0.01–80 Hz band-pass; b1 μV peak to peak noise; 500 Hz samplingrate). These technological advances offer exciting prospects from aresearch as well as a business perspective.

4. Bridging the gap

As outlined above, a range of mobile EEG systems are availableon the market. However, in our opinion none of them meets allthe qualities required for neuroscience experiments in the field in anoptimal way. For this reason a refined mobile and wireless system, theOldenburg EEG system, was recently developed by Debener andcolleagues, which is based on the Emotiv EPOC described above. Theaim of the development was to create a system allowing flexible yetaccurate electrode positioning and good data quality. To achieve this,the original Emotiv EPOC hardware was relocated into a small andlight plastic box (49×44×25mm; 48 g total weight), tightly attachedto the back of a state-of-the-art electrode cap (see Fig. 1) to create ahead-mounted amplifier. Moreover, to improve EEG signal quality theoriginal saline sensors of the Emotiv EPOCwere replaced by 14 sintered

Ag/AgCl electrodes that can be easily connected to the amplifier.The electrodes can virtually be positioned anywhere on a standard EEGcap (www.easycap,de), allowing for a flexible electrode layout (cf.Fig. 1). This is important as it enables to place electrodes over targetedregions best suited for a particular research question or in accordancewith individual differences in the scalp distribution of neural signals.Pilot work conducted by the Oldenburg Neuropsychology lab suggeststhat the Oldenburg EEG system can be used to conduct auditory oddball,P300 and motor imagery studies in different recording environments(personal communication with Professor Debener). In these studiesthe EEG system has been used either in combination with the softwareOpenVibe (Renard et al., 2010) or the software BCI 2000 (Schalk andMellinger, 2010).

In the first published study using the Oldenburg EEG system,Debener et al. (2012) recorded data from an auditory oddball paradigmboth indoors, while participants were sitting in an office, and outdoors,while participants were naturally walking aroundOldenburg Universitycampus. EEG data quality and ERPs of the two recording conditionswere compared. The authors report a strong test–retest associationbetween indoor and outdoor conditions and thereby showed that itis possible to reliably record a cognitive ERP like the P300 in a naturalsetting. Yet open questions remain. For instance, the study also reportsthat the outside P300 had a smaller amplitude, and that outsiderecordings resulted in significantly lower single trial classificationaccuracies (indoor: 77%; outdoor: 69%). This could be explained bya lower signal to noise ratio or different cognitive demands in theoutside condition. Studies are currently under way to help clarifyingthis question.

Studies investigating the brain in its natural state represent a majorstep forward in mobile EEG research, as they demonstrates that brainactivity can be recorded reliably in the real world with mobile EEG,and that mobile EEG is more than just a fancy gadget. In addition,in our opinion mobile EEG is almost certain to become a life changingresource for persons with limited physical mobility or compromisedbrain function. For instance, it is easy to envisage mobile EEG asa diagnostic tool, linking certain brain states or abnormalities in brainactivity to neurological or psychiatric symptoms in a much moreaccessible fashion then presently available. This would not only drivedown health care costs but also, and most critically speed up thediagnosis. Following the same line of argumentation, mobile EEGcould further be used to monitor brain activity in outpatients to ensuretreatment success, continued symptom control, or to indicate that achange in treatment might be required. This would be particularlyinteresting for patient groups finding it difficult to access healthservices, e.g. those living in remote areas or those having difficultytravelling safely by themselves.Mobile EEG is also likely to revolutionizethe use of EEG signals for the replacement of lost abilities, such as theP300 speller if normal channels for communication cannot be accessed,or mental imagery for the control of an electric wheelchair if a patient isnot able to generate a motor output for that purpose. Another likelyapplication of mobile EEG is that of an easy-to-use treatment tool inthe neurofeedback-based therapy of psychiatric disorders, or of motorimpairments resulting from neurological conditions such as strokeor Parkinson's disease.

5. Using mobile EEG to study motor imagery

5.1. Theoretical background: the neurosimulation theory of action

Jeannerod's theory of motor cognition (Jeannerod, 2001) suggeststhe neural simulation of actions as a key mechanism for motor control.According to this theory, action representation comprises overt as well ascovert stages. The overt stage refers to the actual execution ofmovementsand as such represents the stagemost commonly associatedwith actions.However,motor cognition also comprises a number of covert stageswhichrefer to e.g. intended movements, imagination of movements, as well as

Fig. 1. The Oldenburg EEG system in use. Left: Rear view of a participant wearing the systemwhile performing a P300 speller in the laboratory. The amplifier is fixed to the lower edge ofthe EEG cap. The cap is fitted with an electrode layout optimized for P300 recordings. Middle: Participant equipped to perform an auditory oddball experiment while walking outdoors.Data are transmitted wirelessly via bluetooth to a laptop normally carried in a backpack by the participant. Right: Participant performing a motor imagery task outside the laboratory.Electrodes are placed in accordance to the scalp distribution of the sensory motor rhythm of the participant.

13C. Kranczioch et al. / International Journal of Psychophysiology 91 (2014) 10–15

the understanding of actions performed by others. Specifically, Jeannerodtheorized that the motor system forms part of a neurosimulationnetwork which can be activated under different conditions. Critically,these activation conditions do not rely on execution, but neverthelessinvolve similar neural representations. The neurosimulation theorytherefore provides an attractive concept to (1) study characteristicsof complex motor acts, and (2) re-train motor representations via theindirect route of MI or movement observation.

5.2. Motor imagery — motor cortex activation without execution

One particularly interesting area of research in covert actionrepresentation relates to motor imagery (MI). MI refers to the cognitiveprocess throughwhich themental representationof an action is activated.In essence this cognitive process represents ‘motor activation withoutexecution.’ For example studies have shown that kinesthetic MIinvolves the same neural network as motor planning (Jeannerod, 1994;Jeannerod and Frak, 1999) which in turn is thought to rely on the samemotor structures as motor execution (Johnson-Frey, 2004; Munzertet al., 2009; Sharma et al., 2006; Malouin et al., 2007; Malouin &Richards, 2010). In support of this view,MI shares a number of similaritieswith overt movement execution, such as behavioral characteristics(Jeannerod, 1994), physiological parameters (Kranczioch et al., 2009,2010; Liepert et al., 2012), as well as functional neuroanatomicalcorrelates (Decety, 1996b; Lotze and Halsband, 2006; Porro et al., 1996;Szameitat et al., 2007a, 2007b; Hanakawa et al., 2008; Iseki et al., 2008;Palmiero et al., 2009; Ueno et al., 2010; Szameitat et al., 2012). Moreover,mental practice of MI can enhance motor performance and learning(Dickstein and Deutsch, 2007; Mulder, 2007; Braun et al., 2008; Verbuntet al., 2008; Malouin et al., 2009; Malouin & Richards, 2010), and hasbeen applied with relative success in sports training and neurologicalrehabilitation (Braun et al., 2006; Cramer et al., 2007; Garrison et al.,2010; Jackson et al., 2001, 2003; Johnson-Frey, 2004; Lafleur et al.,2002; Page, 2001; Page et al., 2009).

Most MI paradigms use the repetitive imagination of motion or acombination of movements over a period of time. Different types ofMI are thereby distinguished: visual vs. kinesthetic MI (Dickstein andDeutsch, 2007; Malouin et al., 2007), performed in First or Third personperspective (Mulder, 2007; Schuster et al., 2011). In visual imagery,MI is focused on the mental image of the movement, i.e. what amovement looks like. Kinesthetic MI, on the other hand, focuses onthe proprioceptive aspects of themovement, i.e. how it feels to perform

a particularmovement. In First personMI, this imagination is performedas oneself doing themovement, while in Third personMI, one imaginesto watch another person doing the movement. Generally, kinestheticimagery performed in First person perspective induces the most robustactivation in the motor system (e.g. Mulder, 2007).

In the applied/rehabilitation context, MI can be used as a stand-alone intervention, added to other forms of practice or embedded intoanother form of intervention. In the added condition, MI is a standaloneintervention, provided alongside traditional training or rehabilitationtechniques. In the embedded condition, elements of MI are interleavedwith actual movement execution and motor practice (Schuster et al.,2012).

Research onwhether additionalMI training improvesmotor recoveryhas provided inconclusive results. For example, studies focusing onmotor rehabilitation after stroke do report beneficial effects (Liu et al.,2004; Page et al., 2007; Zimmermann-Schlatter et al., 2008) as wellas no additional improvement (Ietswaart et al., 2011). An explanationfor the conflicting results is the variety of MI paradigms and a lack ofevidence-based guidelines in this field. Moreover, since MI is a covertaction, it does not allow judging patients' compliance with instructionsand training protocols.

5.3. EEG signatures of motor imagery

Activity duringMI hasmostly been studied withMRI-based imagingmethods. However, EEG has also been used to successfully examine theinformation processing associated withMI. For example, the contingentnegative variation (CNV) and the lateralized readiness potential (LRP),derived from high-density EEG recordings, have been used to studythe functional similarity of advanced movement preparation in motorexecution and MI (Caldara et al., 2004; Cunnington et al., 1996;Jankelowitz and Colebatch, 2002; Kranczioch et al., 2009, 2010). Resultsfrom these studies suggest that preparatory ERP waveforms forimagined and executed movements are reasonably similar in the earlypreparatory phase but show reduced amplitudes for imaginedmovements in the later stages of the preparatory phase. The latterpresumably results from reduced activation of the primary motor cortex(M1) in imagery, which is in line with the idea that the key differencebetween MI and execution arises from the activation/inhibition ofprimary motor neurons and the corticospinal tract.

Another EEG-based methodology for studying neural correlates ofMI concerns the analysis of activity in the frequency domain. The so-

14 C. Kranczioch et al. / International Journal of Psychophysiology 91 (2014) 10–15

called μ- or wicket-rhythm is characterized by activity in the 8–13 Hzrange, and localized predominantly over the postcentral somatosensorycortex (Pfurtscheller and Lopes da Silva, 1999). Typically, additionalphase-synchronized activity can be found in the β-band (13–30 Hz)over the precentral motor cortex (Nikulin and Brismar, 2006). As bothtypes of activity originate in the sensorimotor cortex, they are referredto as sensorimotor rhythms (SMRs) (Pfurtscheller and Neuper, 2001).Cortical activity leads to a loss of EEG synchrony, which is accompaniedby a suppression of SMR amplitude, a phenomenon also known as event-related desynchronization (ERD). A modulation of SMRs can result fromexternal events like somatosensory stimulation (Nikouline et al., 2000).They can also be modulated by voluntary internal drives, such as theexecution or imagination of movements (Decety, 1996a; Jasper, 1949;Pfurtscheller and Aranibar, 1979; Pfurtscheller et al., 1997). Moreover,the ERD patterns for executed and imagined movements in the μ- andβ-frequency bands are very similar (McFarland et al., 2000). This andthe fact that most people are able to voluntarily control their SMRamplitude using MI as mental strategy opens the door for EEG-basedBCI applications in the field of neurorehabilitation (e.g. Prasad et al.,2010; Cincotti et al., 2012; Ortner et al., 2012).

5.4. Neurofeedback training of motor imagery in healthy personsand patients

The idea of covertly training the motor system through MI isintriguing and promising at the same time. However, the evidence-base for the effects of MI training on brain plasticity and motorperformance is relatively mixed. This is in part has something to dowith the fact that intervention studies are often on a smaller scale andnot always managed to control for confounding variables such as lesionlocation, level of impairment and chronicity. This adds noise to the dataandmight verywell weaken the evidence-base.Moreover, asmentionedabove, because MI is a covert action, it is a stimulation method that doesnot really afford experimental control. Given that the ability to performMI is quite varied across participants, the lack of experimental controlin MI studies is a problem.

EEG-based neurofeedback training enables participants to learnto generate motor cortex activity through imagination in a reliableand measurable way, which solves the problem of poor experimentalcontrol in two ways. First, participants can be trained to criterion toensure they are all able to perform the MI task to an appropriate level.Second, the actual task performance can be monitored and loggedwith EEG recordings. Both aspects are highly important for basic studieson MI, and of course for MI-based training interventions. Moreover,the immediacy of real-time feedback greatly facilitates learning in allparticipants, but probably even more so in poorer learners, not in theleast because it helps to focus and sustain attention, and instigates asense of control and agency. Together these factors enhancemotivationand hence support compliance with a training regime.

To the best of our knowledge only few studies so far have exploredneurofeedback training of MI. Shindo et al. (2011) showed that strokepatients were much more efficient in using a mechanical hand orthosisif they previously received EEG-based neurofeedback training tohelp them master MI. A recent study by Mihara et al. (2013) usednear infrared spectroscopy-mediated (NIRS) neurofeedback to improveMI efficacy in patients with stroke. In a study with 20 hemipareticpatients they found that neurofeedback training enhanced the efficacyof MI-based rehabilitation. Kaiser et al. (2012) compared indices ofhemispheric asymmetry in 29 patients with upper limb hemiparesisderived fromEEG-based neurofeedback duringMI andmotor execution.The authors showed that the asymmetry index derived from MI wassimilar to that of motor execution. Moreover, they could demonstratethat greater asymmetry correlated with poorer motor outcome. Thisis not in itself a new finding. However, the novelty of this study lieson the neurofeedback technology: if it is possible to derive indices ofhemispheric asymmetry through neurofeedback-assisted MI, then it

should be possible to use the same technological approach as a BCI toretrain the brain towards a more symmetrical organization of themotor system which in turn should improve recovery.

The studies summarized above identify EEG-based neurofeedbacktraining as a promising approach for neurorehabilitation and BCIs.With affordablemobile EEG systems capable of providingneurofeedbacktraining in everyday settings, such as the home, it is possible to harnessthe potential of MI-based rehabilitation in full and to build supportivetreatment strategies around it. Through the continued feedback providedon the MI task patients can not only learn to increase the efficacy of MI,but may also experience a sense of control which in turn promotesmotivation and treatment adherence.

6. Conclusion

Mobile EEG is an exciting new technology with excellent potentialfor translational and applied research. However, to fulfill this potentialEEG systems must not only be small and suitable for everyday settings,but also need to produce the data quality required for single trial analysis.One particularly exciting application of mobile EEG is neurofeedbacktraining for MI and/or motor-imagery based BCIs. Initial evidencesuggests that neurofeedback can increase MI effectiveness and helpspatients to learn effective MI strategies. At the same time suchknowledge will help to better characterize motor control and actionrepresentation and as such aids to refine theoretical approaches as wellas clinical applications.

Acknowledgements

CK is supported by grant KR 3433/2-1, German Research Foundation(DFG).

References

Badcock, N.A., M.P, Mahajan, Y., de Lissa, P., Thie, J., McArthur, G., 2013. Validation of theEmotiv EPOC(®) EEG gaming system for measuring research quality auditory ERPs.PeerJ 1, e38. http://dx.doi.org/10.7717/peerj.38.

Braun, S.M., Beurskens, A.J., Borm, P.J., Schack, T., Wade, D.T., 2006. The effects of mentalpractice in stroke rehabilitation: a systematic review. Arch. Phys. Med. Rehabil. 87(6), 842–852. http://dx.doi.org/10.1016/j.apmr.2006.02.034.

Braun, S., Kleynen, M., Schols, J., Schack, T., Beurskens, A., Wade, D., 2008. Using mentalpractice in stroke rehabilitation: a framework. Clin. Rehabil. 22 (7), 579–591.

Caldara, R., Deiber, M.P., Andrey, C., Michel, C.M., Thut, G., Hauert, C., 2004. Actualand mental motor preparation and execution: a spatiotemporal ERP study. Exp. BrainRes. 159, 389–399.

Cincotti, F., Pichiorri, F., Arico, P., Aloise, F., Leotta, F., de Vico Fallani, F., Mattia, D.,2012. EEG-based brain–computer interface to support post-stroke motorrehabilitation of the upper limb. Conf. Proc. IEEE Eng. Med. Biol. Soc. 2012, 4112–4115.http://dx.doi.org/10.1109/EMBC.2012.6346871.

Cramer, S.C., Orr, E.L., Cohen, M.J., Lacourse, M.G., 2007. Effects of motor imagerytraining after chronic, complete spinal cord injury. Exp. Brain Res. 177 (2), 233–242.http://dx.doi.org/10.1007/s00221-006-0662-9.

Cunnington, R., Iansek, R., Bradshaw, J.L., Phillips, J.G., 1996. Movement-related potentialsassociated with movement preparation and motor imagery. Exp. Brain Res. 111 (3),429–436.

De Sanctis, P., Butler, J.S., Green, J.M., Snyder, A.C., Foxe, J.J., 2012. Mobile brain/bodyimaging (MoBi): high-density electrical mapping of inhibitory processes duringwalking. Conf. Proc. IEEE Eng. Med. Biol. Soc. 2012, 1542–1545.

Debener, S., Minow, F., Emkes, R., Gandras, K., De Vos, M., 2012. How about taking a low-cost, small, and wireless EEG for a walk? Psychophysiology 49 (11), 1449–1453.http://dx.doi.org/10.1111/j.1469-8986.2012.01471.x.

Decety, J., 1996a. Do imagined and executed actions share the same neural substrate?Brain Res. Cogn. Brain Res. 3 (2), 87–93.

Decety, J., 1996b. The neurophysiological basis of motor imagery. Behav. Brain Res. 77(1–2), 45–52.

Dickstein, R., Deutsch, J.E., 2007. Motor imagery in physical therapist practice. Phys. Ther.87 (7), 942–953. http://dx.doi.org/10.2522/ptj.20060331.

Garrison, K.A., Winstein, C.J., Aziz-Zadeh, L., 2010. The mirror neuron system: a neuralsubstrate for methods in stroke rehabilitation. Neurorehabil. Neural Repair 24 (5),404–412. http://dx.doi.org/10.1177/1545968309354536.

Gramann, K., Gwin, J.T., Bigdely-Shamlo, N., Ferris, D.P., Makeig, S., 2010. Visual evokedresponses during standing and walking. Front. Hum. Neurosci. 4, 202.

Gramann, K., Gwin, J.T., Ferris, D.P., Oie, K., Jung, T.P., Lin, C.T., Liao, L.D., Makeig, S., 2011.Cognition in action: imaging brain/body dynamics in mobile humans. Rev. Neurosci.22, 593–608.

15C. Kranczioch et al. / International Journal of Psychophysiology 91 (2014) 10–15

Hanakawa, T., Dimyan, M.A., Hallett, M., 2008. Motor planning, imagery, and executionin the distributed motor network: a time-course study with functional MRI. Cereb.Cortex. http://dx.doi.org/10.1093/cercor/bhn036.

Ietswaart, M., Johnston, M., Dijkerman, H.C., Joice, S., Scott, C.L., MacWalter, R.S., Hamilton,S.J., 2011. Mental practice with motor imagery in stroke recovery: randomizedcontrolled trial of efficacy. Brain 134 (Pt 5), 1373–1386. http://dx.doi.org/10.1093/brain/awr077.

Iseki, K., Hanakawaa, T., Shinozakia, J., Nankakuc, M., Fukuyamaa, H., 2008. Neuralmechanisms involved in mental imagery and observation of gait. NeuroImage 41 (3),1021–1031.

Jackson, P.L., Lafleur,M.F., Malouin, F., Richards, C., Doyon, J., 2001. Potential role of mentalpractice using motor imagery in neurologic rehabilitation. Arch. Phys. Med. Rehabil.82 (8), 1133–1141. http://dx.doi.org/10.1053/apmr.2001.24286.

Jackson, P.L., Lafleur, M.F., Malouin, F., Richards, C.L., Doyon, J., 2003. Functional cerebralreorganization following motor sequence learning through mental practice withmotor imagery. NeuroImage 20 (2), 1171–1180.

Jankelowitz, S.K., Colebatch, J.G., 2002. Movement-related potentials associated with self-paced, cued and imagined arm movements. Exp. Brain Res. 147 (1), 98–107.

Jasper, H.P.W., 1949. Electrocorticograms in man: effect of voluntary movement upon theelectrical activity of the precentral gyrus. Arch. Psychiatr. Nervenkr. 183 (1–2), 163–174.

Jeannerod, M., 1994. The representing brain: neural correlates of motor intention andimagery. Behav. Brain Sci. 17 (2).

Jeannerod, M., 2001. Neural simulation of action: a unifying mechanism for motorcognition. NeuroImage 14 (1), S103–S109.

Jeannerod, M., Frak, V., 1999. Mental imaging of motor activity in humans. Curr. Opin.Neurobiol. 9 (6), 735–739.

Johnson-Frey, S.H., 2004. Stimulation through simulation? Motor imagery andfunctional reorganization in hemiplegic stroke patients. Brain Cogn. 55 (2),328–331. http://dx.doi.org/10.1016/j.bandc.2004.02.032.

Kaiser, V., Daly, I., Pichiorri, F., Mattia, D., Muller-Putz, G.R., Neuper, C., 2012. Relationshipbetween electrical brain responses to motor imagery and motor impairment in stroke.Stroke 43 (10), 2735–2740. http://dx.doi.org/10.1161/STROKEAHA.112.665489.

Kranczioch, C., Mathews, S., Dean, P.J., Sterr, A., 2009. On the equivalence of executed andimagined movements: evidence from lateralized motor and nonmotor potentials.Hum. Brain Mapp. http://dx.doi.org/10.1002/hbm.20748.

Kranczioch, C., Mathews, S., Dean, P.J.A., Sterr, A., 2010. Task complexity differentiallyaffects executed and imagined movement preparation. Evidence from movement-related potentials. PLoS ONE 5(2). http://dx.doi.org/10.1371/journal.pone.0009284.

Lafleur, M.F., Jackson, P.L., Malouin, F., Richards, C.L., Evans, A.C., Doyon, J., 2002.Motor learning produces parallel dynamic functional changes during the executionand imagination of sequential foot movements. NeuroImage 16 (1), 142–157.http://dx.doi.org/10.1006/nimg.2001.1048.

Liepert, J., Greiner, J., Nedelko, V., Dettmers, C., 2012. Reduced upper limb sensation impairsmental chronometry for motor imagery after stroke: clinical and electrophysiologicalfindings. Neurorehabil. Neural Repair 26 (5), 470–478. http://dx.doi.org/10.1177/1545968311425924.

Liu, K.P., Chan, C.C., Lee, T.M., Hui-Chan, C.W., 2004. Mental imagery for promotingrelearning for people after stroke: a randomized controlled trial. Arch. Phys. Med.Rehabil. 85 (9), 1403–1408.

Lotze, M., Halsband, U., 2006. Motor imagery. J. Physiol. Paris 99 (4–6), 386–395.http://dx.doi.org/10.1016/j.jphysparis.2006.03.012.

Makeig, S., Gramann, K., Jung, T.P., Sejnowski, T.J., Poizner, H., 2009. Linking brain,mind and behavior. Int. J. Psychophysiol. 73, 95–100.

Malouin, F., Richards, C.L., Jackson, P.L., Lafleur, M.F., Durand, A., Doyon, J., 2007.The Kinesthetic and Visual Imagery Questionnaire (KVIQ) for assessing motor imageryin persons with physical disabilities: a reliability and construct validity study. J. Neurol.Phys. Ther. 31 (1), 20–29.

Malouin, F., Richards, C.L., Durand, A., Doyon, J., 2009. Added value of mentalpractice combined with a small amount of physical practice on the relearning ofrising and sitting post-stroke: a pilot study. J. Neurol. Phys. Ther. 33 (4), 195–202.http://dx.doi.org/10.1097/NPT.0b013e3181c2112b.

Malouin, F., Richards, C.L., 2010. Mental practice for relearning locomotor skills. Phys.Ther. 90, 240–251. http://dx.doi.org/10.2522/ptj.20090029.

McFarland, D.J., Miner, L.A., Vaughan, T.M., Wolpaw, J.R., 2000. Mu and beta rhythmtopographies during motor imagery and actual movements. Brain Topogr. 12 (3),177–186.

Mihara, M., Hattori, N., Hatakenaka, M., Yagura, H., Kawano, T., Hino, T., Miyai, I., 2013.Near-infrared spectroscopy-mediated neurofeedback enhances efficacy of motorimagery-based training in poststroke victims: a pilot study. Stroke 44 (4),1091–1098. http://dx.doi.org/10.1161/STROKEAHA.111.674507.

Mulder, T., 2007. Motor imagery and action observation: cognitive tools for rehabilitation.J. Neural Transm. 114 (10), 1265–1278. http://dx.doi.org/10.1007/s00702-007-0763-z.

Munzert, J., Lorey, B., Zentgraf, K., 2009. Cognitive motor processes: the role of motorimagery in the study of motor representations. Brain Res. Rev. 60 (2), 306–326.http://dx.doi.org/10.1016/j.brainresrev.2008.12.024.

Nikouline, V.V., Linkenkaer-Hansen, K., Wikstrom, H., Kesaniemi, M., Antonova,E.V., Ilmoniemi, R.J., Huttunen, J., 2000. Dynamics of mu-rhythm suppressioncaused by median nerve stimulation: a magnetoencephalographic study inhuman subjects. Neurosci. Lett. 294 (3), 163–166.

Nikulin, V.V., Brismar, T., 2006. Phase synchronization between alpha and betaoscillations in the human electroencephalogram. Neuroscience 137 (2), 647–657.http://dx.doi.org/10.1016/j.neuroscience.2005.10.031.

Ortner, R., Irimia, D.C., Scharinger, J., Guger, C., 2012. Amotor imagery based brain–computerinterface for stroke rehabilitation. Stud. Health Technol. Inform. 181, 319–323.

Page, S.J., 2001. Mental practice: a promising restorative technique in strokerehabilitation. Top. Stroke Rehabil. 8 (3), 54–63.

Page, S.J., Levine, P., Leonard, A., 2007. Mental practice in chronic stroke: results of arandomized, placebo-controlled trial. Stroke 38 (4), 1293–1297. http://dx.doi.org/10.1161/01.STR.0000260205.67348.2b.

Page, S.J., Szaflarski, J.P., Eliassen, J.C., Pan, H., Cramer, S.C., 2009. Cortical plasticityfollowing motor skill learning during mental practice in stroke. Neurorehabil. NeuralRepair 23 (4), 382–388. http://dx.doi.org/10.1177/1545968308326427.

Palmiero, M., Belardinelli, M.O., Nardo, D., Sestieri, C., DiMatteo, R., D'Ausilio, A., Romani, G.L.,2009. Mental imagery generation in different modalities activates sensory-motor areas.Cogn. Process. 10 (Suppl. 2), S268–S271. http://dx.doi.org/10.1007/s10339-009-0324-5.

Pfurtscheller, G., Aranibar, A., 1979. Evaluation of event-related desynchronization (ERD)preceding and following voluntary self-paced movement. Electroencephalogr. Clin.Neurophysiol. 46 (2), 138–146.

Pfurtscheller, G., Lopes da Silva, F.H., 1999. Event-related EEG/MEG synchronization anddesynchronization: basic principles. Clin. Neurophysiol. 110 (11), 1842–1857.

Pfurtscheller, G., Neuper, C., 2001. Motor imagery and direct brain–computercommunication. Proc. IEEE 89 (7), 1123–1134.

Pfurtscheller, G., Neuper, C., Flotzinger, D., Pregenzer, M., 1997. EEG-based discriminationbetween imagination of right and left hand movement. Electroencephalogr. Clin.Neurophysiol. 103 (6), 642–651.

Porro, C.A., Francescato, M.P., Cettolo, V., Diamond, M.E., Baraldi, P., Zuiani, C., diPrampero, P.E., 1996. Primary motor and sensory cortex activation during motorperformance and motor imagery: a functional magnetic resonance imaging study.J. Neurosci. 16 (23), 7688–7698.

Prasad, G., Herman, P., Coyle, D., McDonough, S., Crosbie, J., 2010. Applying a brain–computer interface to support motor imagery practice in people withstroke for upper limb recovery: a feasibility study. J. Neuroeng. Rehabil. 7, 60.http://dx.doi.org/10.1186/1743-0003-7-60.

Renard, Y., Lotte, F., Gibert, G., Congedo, M., Delannoy, V., Bertrand, O., Lécuyer, A., 2010.OpenViBE: an open-source software platform to design, test, and use brain–computerinterfaces in real and virtual environments. Presence Teleoper. Virtual Environ. 19(1), 35–53.

Schalk, G., Mellinger, J., 2010. A Practical Guide to Brain–Computer Interfacing WithBCI2000. Springer, London, UK.

Schuster, C., Hilfiker, R., Amft, O., Scheidhauer, A., Andrews, B., Butler, J., Ettlin, T., 2011.Best practice for motor imagery: a systematic literature review on motor imagerytraining elements in five different disciplines. BMC Med. 9, 75. http://dx.doi.org/10.1186/1741-7015-9-75.

Schuster, C., Butler, J., Andrews, B., Kischka, U., Ettlin, T., 2012. Comparison of embeddedand added motor imagery training in patients after stroke: results of a randomisedcontrolled pilot trial. Trials 13, 11. http://dx.doi.org/10.1186/1745-6215-13-11.

Sharma, N., Pomeroy, V.M., Baron, J.C., 2006. Motor imagery: a backdoor to the motorsystem after stroke? Stroke 37, 1941–1952.

Shindo, K., Kawashima, K., Ushiba, J., Ota, N., Ito, M., Ota, T., Liu, M., 2011. Effectsof neurofeedback training with an electroencephalogram-based brain–computerinterface for hand paralysis in patients with chronic stroke: a preliminary case seriesstudy. J. Rehabil. Med. 43 (10), 951–957. http://dx.doi.org/10.2340/16501977-0859.

Szameitat, A.J., Shen, S., Sterr, A., 2007a. Effector-dependent activity in the left dorsalpremotor cortex in motor imagery. Eur. J. Neurosci. 26 (11), 3303–3308.

Szameitat, A.J., Shen, S., Sterr, A., 2007b. Motor imagery of complex everyday movements.An fMRI study. NeuroImage 34 (2), 702–713.

Szameitat, A.J., Shen, S., Conforto, A., Sterr, A., 2012. Cortical activation duringexecuted, imagined, observed, and passive wrist movements in healthyvolunteers and stroke patients. NeuroImage 62 (1), 266–280. http://dx.doi.org/10.1016/j.neuroimage.2012.05.009.

Ueno, T., Inoue,M., Matsuoka, T., Abe, T., Maeda, H., Morita, K., 2010. Comparison betweena real sequential finger and imagery movements: an FMRI study revisited. BrainImaging Behav. 4 (1), 80–85.

Verbunt, J.A., Seelen, H.A., Ramos, F.P., Michielsen, B.H., Wetzelaer, W.L., Moennekens, M.,2008. Mental practice-based rehabilitation training to improve arm function anddaily activity performance in stroke patients: a randomized clinical trial. BMC Neurol.8, 7. http://dx.doi.org/10.1186/1471-2377-8-7.

Zimmermann-Schlatter, A., Schuster, C., Puhan, M.A., Siekierka, E., Steurer, J., 2008.Efficacy of motor imagery in post-stroke rehabilitation: a systematic review.J. Neuroeng. Rehabil. 5, 8. http://dx.doi.org/10.1186/1743-0003-5-8.