CASE REPORT

Non-invasive pre-surgical investigation of a10 year-old epileptic boy using simultaneousEEG—NIRS

Anne Gallagher a,b, Maryse Lassonde a,b, Danielle Bastien a,b,Phetsamone Vannasing b, Frederic Lesage c, Christophe Grova d,Alain Bouthillier e, Lionel Carmant b,f, Franco Lepore a,b,Renee Beland a, Dang Khoa Nguyen e,*

aCentre de Recherche en Neuropsychologie et Cognition, Universite de Montreal,Departement de Psychologie, C.P. 6128, Succ. Centre-Ville, Montreal, Quebec,H3C 3J7, CanadabCentre de Recherche de l’Hopital Sainte-Justine, Hopital Sainte-Justine,3175 Chemin de la Cote Sainte-Catherine, Montreal, Quebec, H3T 1C5, Canadac Ecole Polytechnique, Universite de Montreal, C.P. 6079, Succ. Centre-ville, Montreal,Quebec, H3C 3A7, Canadad Institut Neurologique de Montreal, 3801 rue Universite, Montreal, Quebec, H3A 2B4, Canadae Service de Neurologie, Hopital Notre-Dame du CHUM, 1560 rue Sherbrooke Est, Montreal,Quebec, H2L 4M1, Canadaf Service de Neurologie, Hopital Sainte-Justine, 3175 Chemin de la Cote Sainte-Catherine,Montreal, H3T 1C5, Quebec, Canada

Received 8 November 2007; accepted 23 January 2008

Seizure (2008) 17, 576—582

www.elsevier.com/locate/yseiz

KEYWORDSFrontal lobe epilepsy;Pre-surgicalassessment;Near-infraredspectroscopy (NIRS);Electrophysiology(EEG);Surgical outcome;Children

Summary Near-infrared spectroscopy (NIRS) is a novel imaging technique ofpotential value in the pre-surgical investigation of patients with refractory epi-lepsy. We recorded simultaneously electrophysiology (EEG; Compumedics, USA)and near-infrared spectroscopy (NIRS; ISS, USA) to examine the localization of theictal onset zone and assess language lateralization in a young epileptic boy (L.H.,10 years) as part of his pre-surgical evaluation. L.H. underwent a prolonged EEG—NIRS recording while electro-clinical and electrical seizures were recorded. Resultswere compared to those obtained with other pre-surgical techniques (SPECT, FDG—PET, EEG—fMRI and EEG—MEG) and showed good concordance for ictal onset zonelocalization. A second NIRS session without EEG was carried out in order toinvestigate language lateralization. For this purpose, the patient performed a

categorical verbal-fluency task during NIRS recordings. Results showed left-hemi-

* Corresponding author. Tel.: +1 514 890 8237; fax: +1 514 412 7554.E-mail address: d.nguyen@umontreal.ca (D.K. Nguyen).

1059-1311/$ — see front matter # 2008 British Epilepsy Association. Published by Elsevier Ltd. All rights reserved.doi:10.1016/j.seizure.2008.01.009

Non-invasive pre-surgical investigation of an epileptic boy 577

sphere dominance for language function in this young boy. This case reportillustrates that multi-channel EEG—NIRS has the potential to contribute favourablyto pre-surgical investigation in young patients.# 2008 British Epilepsy Association. Published by Elsevier Ltd. All rights reserved.

Introduction

Localization of the ictal onset zone remains a chal-lenge in patients with non-lesional neocortical epi-lepsy. When the magnetic resonance imaging (MRI) isnormal, one must typically rely on the clinical andneuropsychological evaluations, video-electroence-phalography (EEG) monitoring, inter-ictal/ictal sin-gle-photon emission computed tomography (SPECT),and positron-emission tomography (PET).1 Whenthese complementary assessments fail to adequatelylocalize the ictal onset zone, invasive intracranialelectrode studies are required but carry a risk ofinfection and haemorrhage.2 In some cases, languagelateralization investigation is also required and it isusually done using the invasive intracarotid amobar-bital test (IAT) in conjunction with neuropsychologi-cal testing. In order to reduce the number of caseswhere invasive procedures are necessary, recentimaging techniques such as functional MRI (fMRI)andmagnetoencephalography (MEG) have been usedto investigate language dominance,3,4 and havereplaced the IAT in some clinical centers. Promisingnon-invasive techniques such as simultaneous EEG—fMRI and EEG—MEG have also been considered forlocalizing the ictal onset area.5 However, these tech-niques are sometimes difficult to use with youngchildren and patients with serious cognitive or beha-vioural problems. Alternatively in such cases, near-infrared spectroscopy (NIRS) can non-invasivelymonitor rapid changes of regional cerebral bloodvolume (rCBV) (as seen during seizures or functionalactivations). NIRS is a relatively new technique,which allows the measurement of haemodynamicchanges associated with neural activity.6 The differ-ent light absorption spectra of oxy-haemoglobin(HbO) and deoxy-haemoglobin (HbR) within thenear-infrared spectrum allow for the measurementof concentration changes of these substances in livingtissues.7,8 Near-infrared light of two wavelengthsbetween 680 and 1000 nm is directed through opticfibres to the head of the patient. The amount ofdetected light reflects the amount of absorption ofthe two wavelengths in targeted cerebral areasinforming on concentration changes of HbO andHbR in the regions (for review see Ref. 9).

This technique has several advantages over otherimaging methods.9—11 First, it allows independentmeasurement of concentration changes of HbO and

HbR, as well as measurement of total haemoglobin(HbT), which is the sum of HbO and HbR. Second, theequipment is portable12,13 and less costly than fMRIor MEG. Finally, and most importantly, there are nomajor restrictions on movements or verbalizationduring recording, which renders the technique sui-table for studies in mentally challenged people aswell as in young children, even infants.14 A potentialdisadvantage of this technique is the shallow pene-tration of the photons (between 3 and 5 cm), whichrenders it difficult or impossible to collect reliabledata from subcortical structures. Nonetheless, aspatial resolution below 1 cm can be obtained inmost patients. Furthermore, the limited penetra-tion should not have a major impact on studiesinvestigating cortical areas, especially in childrenwho usually have a thinner skull than adults.

Preliminary NIRS studies with adult epilepticpopulations showed different haemodynamic pat-terns among different seizure types. More pre-cisely, an increase of rCBV, HbT and HbO, hasbeen reported on the side of seizure focus duringcomplex partial seizures,15—17 whereas rapidly sec-ondarily generalized seizures have been associatedwith a decrease of cerebral blood oxygenation.15

NIRS has also been used to investigate languagelateralization in epileptic patients,18 including chil-dren19 showing good concordance with other tech-niques (IAT and fMRI).

Case report

L.H. is a 10 year-old right-handed boy with noseizure risk factor except for an uncle who sufferedfrom refractory seizures. Onset of seizures occurredat 4 years. Seizures were characterized by dailyauras of fear with or without startle, vocalization,left ocular and head deviation. Multiple anti-epi-leptic drugs failed to control seizures. EEG revealeda probable right frontal origin (F4) but no clearlesion was identified on anatomical MRI. Multiplefunctional imaging techniques were also used. Ictal99mTc-ECD SPECT (injection 1 s after seizure onset)showed a major activation in the right superiorfrontal and parasagittal areas with an extensionto the right antero-lateral frontal regions. FDG—PET showed a right frontal hypometabolism. L.H.also underwent testing with novel sophisticated

578 A. Gallagher et al.

techniques (simultaneous EEG—fMRI and simulta-neous EEG—MEG). Simultaneous EEG—fMRI acquisi-tions were used to measure within the whole brainhaemodynamic responses that correlate with epi-leptic discharges detected on scalp EEG (for meth-odological details see Ref. 20). Simultaneous EEG—MEG were used to measure directly on the scalpelectric and magnetic components of signals gener-ated by a population of pyramidal neurons synchro-nously active during an epileptic discharge(temporal resolution–—1 ms, high density coveringall the surface of the head: 275 sensors in a VSM-CTFMEG system and 64 EEG electrodes). MEG cerebralsources have been estimated using the maximunentropy on the mean (MEM) source localizationtechnique.21 Electro-clinical seizures wererecorded during both procedures and marked byan expert electroencephalographer. BOLD responseand MEG source localization showed a clear activa-tion over the right frontal pole during seizure activ-ity. Ictal EEG—fMRI and ictal EEG—MEG were madepossible as the patient had several electrical sei-zures and clinical seizures only accompanied by asensation of fear with little or no movement. Thus,all functional techniques pointed towards a rightfrontal epileptic focus. Paradoxically, neuropsycho-logical data reported language deficits (word findingdifficulties, slowing of verbal information planningand processing), and attention deficits. L.H under-went a prolonged and simultaneous EEG—NIRSrecording in order to evaluate the potential of thisnon-invasive technique to improve the localizationof the ictal onset zone. A second NIRS recordingwithout EEG was done to assess language lateraliza-tion. The project was previously approved by theEthics Committees of the Sainte-Justine and Notre-Dame Hospitals.

Methods

L.H. underwent a 3-h EEG—NIRS video-monitoringsession during which electro-clinical and electricalseizures were recorded. During recording, thepatient remained passive or slept. One hundredand twenty-eight-NIRS channels placed over rightfrontal, bilateral parasagittal regions and bilateralrolandic regions were recorded using a multi-chan-nel Imagent Tissue Oxymeter (ISS, USA). Opticalintensity (DC), modulation amplitude (AC) andphase data were sampled at 39.0625 Hz. DC datawere filtered using a band-pass filter of 0.1—0.001 Hz, normalized by dividing each value bythe mean value across time points and transformedto quantify concentration changes of HbO and HbRfor each channel. Standard EEG video-monitoring

was carried out simultaneously using 18 home-madecarbon fibre electrodes and a NeuroScan Synamps2TM system (Compumedics, USA). EEG data wereacquired using a sampling rate of 500 Hz with aband-pass filter of 0.1—100 Hz and a central refer-ence. Electrodes were placed based on the 10—20system (Fz, Cz, Pz, Oz, Fp1, F3, C3, P3, O1, Fp2, F4,C4, P4, O2, VeOg, HeOg, EKG and EMG). Epilepticactivity on EEG and clinical manifestations recordedon video were used to assess onset and duration ofseizures. Epileptic discharges were manuallymarked by an expert electroencephalographer.Optical fibres and electrodes were placed on thesurface of the head, using a light and comfortable,but rigid, helmet, which can be adapted to all headsizes without restricting head movements.

In order to investigate language dominance, asecond NIRS recording without EEG was carried outusing 128-NIRS channels that covered left Broca’sand Wernicke’s areas and right homologous areas.During this session, NIRS data were acquired in ablock design paradigm and were then averaged byblocks (10 for each task). A categorical verbal-flu-ency task (language task) and a non-sense syllablerepetition task (oro-motor control task) were per-formed by L.H. during this second NIRS recording.

For each NIRS session, an anatomically specificmontage was created based on the MRI of thepatient to ensure that optical fibres were placedover the regions of interest (ROI) specific to theobjectives of each recording session (session 1: rightfrontal, bilateral parasagittal areas and bilateralrolandic regions; session 2: left Broca’s and Wer-nicke’s areas and right mirror areas). This was doneusing a stereotaxic system (BrainsightTM Frameless39, Rogue Research, Canada), which enables thetransfer of ROI, determined by MRI, onto the hel-met. The location of each optical fibre and fourfiducial points was digitized and recorded using thesame stereotaxic system to allow for precise align-ment of the NIRS and the anatomical data.

Results

Two electro-clinical and two electrical seizureswere recorded over the prolonged EEG—NIRS ses-sion. Electro-clinical seizures lasted longer (30—36 scompared to 4—13 s) and elicited larger haemody-namic responses than electrical seizures (Fig. 1).These seizures were characterized by a suddensensation of fear. Electrically, ictal rhythmic spikesover the right fronto-central region were noted.Ictal NIRS mapping co-registered onto the MRIrevealed a high increase of rCBV over the rightfrontal region during both electro-clinical and elec-

Non-invasive pre-surgical investigation of an epileptic boy 579

Figure 1 EEG—NIRS results during electro-clinical (left) and electrical seizures (right). (a) EEG data during seizures areshown on the top. Simultaneous NIRS data during seizures are shown by (b) the graphs and (c) haemoglobin map images.(b) Graphs show NIRS cerebral activation in five NIRS channels covering right fronto-polar region. Cerebral activation ischaracterized by an enhanced HbO (solid lines) as well as a small and late decrease of HbR (dotted lines), happening a fewseconds after the beginning of the seizure. The horizontal red bar shows the seizure duration. A typical initial dip (shortreduction of HbO) is also obtained before the activation. This haemodynamic signal has been measured on the rightfrontal area (see 3D MRI reconstruction in the middle), which is indicated by the green rectangle. It is important to notethat the scale between both graphs are greatly different (5�) which is showing that cerebral activation is much moreprominent during the electro-clinical seizure compared to the electrical seizure. The importance of this amplitudedifference between both haemodynamic responses did not allow using the same scale. (c) Enhancement of HbTand HbO isshown by haemoglobin map images, where the rectangle is the same as the right frontal rectangle seen on the 3D MRIreconstruction. This activation corresponds temporally to the peak of HbO enhancement.

trical seizures. More specifically, an initial dip (shortreduction of HbO) was measured before an increaseof HbTand HbO, and small decrease of HbR, showinga cerebral activation in this region. This was in goodconcordance with other functional techniques thatwere used in this patient (Fig. 2).

Subsequent NIRS recording was carried out com-paring activations elicited by the language task

(verbal fluency) and the control task (non-sensesyllables repetition task). Temporally, left and rightcerebral activations occurred simultaneouslyaround 30 s after baseline, at the end of the ver-bal-fluency task (Fig. 3c). The latter result suggeststhat some language reorganization may have takenplace in the right hemisphere. Overall, however, amore prominent increase of HbO and decrease of

580 A. Gallagher et al.

Figure 2 Results from SPECT (substraction ictal SPECTco-registered on MRI, SISCOM) (top left), PET (top right), as wellas BOLD response detected using EEG—fMRI (bottom left) and EEG—MEG source (bottom right) for similar EEG dischargesshowed a good concordance between functional neuro-imaging techniques suggesting a right frontal ictal onset zone (redarrows) in this 10 year-old epileptic boy.

HbR were measured in the left inferior—posteriorfrontal gyrus (Broca’s area) compared to its rightanalogous region, indicating a left hemispheric spe-cialization for language (Fig. 3b).

Following a case discussion and examination ofdifferent imaging results, it was decided not toproceed to a long-term invasive video-EEG studybut rather limit ourselves with intraoperative elec-trocorticography immediately followed by surgery.Intraoperative electrocorticography revealed con-tinuous spiking over the suspected right frontalfocus and a limited right frontal corticectomy wasperformed. The patient has remained seizure-freesince the surgery (follow-up 11 months) comparedto multiple daily seizures pre-operatively. Patholo-gical analysis of the resected tissue revealed anunderlying cortical dysplasia.

Discussion

The present work is the first study using 128-NIRSchannels non-invasively during spontaneous sei-zures to localize the ictal onset zone and investi-gate language lateralization in a young epileptic

patient. A clear activation was obtained usingsimultaneous EEG—NIRS in the right frontal regionduring epileptic seizures in this 10 year-old boy,with refractory MRI-negative right frontal epi-lepsy. Adelson et al.22 monitored continuouslycerebral haemoglobin oxygen availability usingNIRS in three patients and Haginoya et al.23 usedNIRS to record various types of epileptic seizuresin 15 children. However, these studies used only afew optodes placed over the patient’s foreheadand were not meant to localize the seizure focus.In another study, Watanabe et al.17 applied NIRSmonitoring to measure bemegride-induced ictalchanges in CBV in 26 (mainly adult) patientsundergoing intracranial recordings using 8—24channels mounted on a thermoplastic splint shellso as to cover the prospective focus region asdetermined by a prior pre-surgical evaluation.Compared to the present study, Watanabeet al.17 used much lesser NIRS channels and mostof the recorded seizures were chemically induced.Surgical outcome was not mentioned.

In the present study, EEG—NIRS results were invery good concordance with the data obtained fromother functional imaging techniques (SPECT, FDG—

Non-invasive pre-surgical investigation of an epileptic boy 581

Figure 3 This figure shows NIRS results during a language task used in order to investigate language lateralization. (a) A3D MRI reconstruction of the patient with the position of the NIRS fibres is shown (sources in red, detectors in yellow).Fibre locations on left Broca’s and Wernicke’s areas and their right homologous regions were determined using astereotaxic system and the patient’s MRI. (b) A clear enhancement of HbO in Broca’s area (left hemisphere) 28 s after thebeginning of the language task suggests a left language lateralization in L.H. A decrease of HbR was also present but notshown here (see c). (c) Graph shows haemoglobin concentration changes during one-block time course including baseline(�30 to 0 s), language task (0—30 s), resting period (30—60 s), control task (60—90 s) and resting period (90—105 s). Left-hemisphere activations are illustrated in blue and right hemisphere activations in pink. HbO concentrations are indicatedas solid lines whereas HbR levels correspond to the dotted lines. Left-hemisphere activation is shown by an increase inHbO concentration and a decrease in HbR concentration occurring at the end of the language task (starting 30 s afterbaseline, see black arrow on the graph, and peaking a few seconds later). This activation is more pronounced than theactivation seen in the right hemisphere (see difference between blue and pink lines between 35 and 70 s, horizontal greenline). A typical initial dip (short reduction of HbO) is also obtained before the activation (black arrow). These data arefrom a central channel in Broca’s area (blue lines) and a central channel in the right homologous region (pink lines).

PET, EEG—fMRI and EEG—MEG) in this young patient.Furthermore, NIRS allowed to assess his languagedominance non-invasively as previously reported byWatanabe et al.18 and our group.19

Conclusion

This case report illustrates that continuous EEG—NIRS has the potential to contribute favourably tothe localization of the ictal onset zone and assess-ment of language lateralization. This recent tech-nique is non-invasive, more resistant to movementartefacts than other techniques and can easilybe performed in young patients. It has a better

temporal resolution than SPECT and fMRI and pro-vides quantitative information about HbT, HbO andHbR compared to fMRI BOLD signal based principallyon HbR variations.

In future studies, it would be interesting to covereven more extensively the whole scalp. For somepatients, careful revision of MRI guided by NIRScould lead to the detection of subtle epileptogeniclesions previously missed by visual inspection. Forothers, EEG—NIRS combined with classical and othernovel non-invasive techniques may reduce the needfor invasive monitoring. And for those who stillrequire an intracranial study, NIRS can potentiallyallowmore accurate electrode positioning or reducethe extent of the studied zone. Moreover, NIRS has

582 A. Gallagher et al.

the potential to become a viable, non-invasivealternative to IAT, fMRI and MEG in the determina-tion of speech lateralization in children and clinicalpopulations who fail to remain motionless or arereluctant to submit to more invasive techniques.

Acknowledgements

We are grateful to Jean Gotman, M.D., Ph.D., LouiseTyvaert, M.D., Eliane Kobayashi, M.D., Ph.D. forproviding EEG—fMRI data, helping in EEG—MEG ana-lyses and their fruitful discussions. We are alsoindebted to the engineering and orthotic prosthesisteam of Sainte-Justine Hospital and Marie-EnfantHospital Center for their help in developing theoptical imaging helmet.

This work was supported by funds from the SavoyFoundation held by Dr. Dang K. Nguyen, the CanadaResearch Chair in Developmental Neuropsychologyheld by Dr. Maryse Lassonde, the Canada ResearchChair in Cognitive Neurosciences held by Dr. FrancoLepore, as well as scholarships by the CanadianInstitutes of Health Research (CIHR), the CanadianFederation of University Women (CFUW), and theFonds de la Recherche en Sante du Quebec (FRSQ)awarded to Anne Gallagher, M.Ps. We confirm thatwe have read the journal’s position on issuesinvolved in ethical publication and affirm that thisreport is consistent with these guidelines.

References

1. Rosenow F, Luders H. Presurgical evaluation of epilepsy. Brain2001;124:1683—700.

2. Nguyen DK, Spencer SS. Invasive EEG evaluation for epilepsysurgery. In: Shorvon S, Fish D, Dodson E, Perucca E, editors.The treatment of epilepsy. 2nd ed. Oxford: Blackwell Pub-lishing; 2004. p. 609—34. [chapter 53].

3. Gaillard WD, Balsamo L, Xu B, McKinney C, Papero PH,Weinstein S, et al. 4 fMRI language task panel improvesdetermination of language dominance. Neurology 2004;63:1403—8.

4. Papanicolaou AC, Simos PG, Castillo EM, Breier JI, Sarkari S,Pataraia E, et al. Magnetocephalography: a non-invasivealternative to the Wada procedure. J Neurosurg 2004;100:867—76.

5. Grova C, Daunizeau J, Kobayashi E, Bagshaw AP, Lina JM,Dubeau F, et al. Concordance between EEG source localiza-tion and simultaneous EEG-fMRI studies of epileptic spikes.NeuroImage 2007;39(2):755—74.

6. Villringer A, Plank J, Hock C, Schleinkofer L, Dirnagl U. Nearinfrared spectroscopy (NIRS): a new tool to studyhemodynamic

changes during activation of brain function in adults. NeurosciLett 1993;154:101—4.

7. Gratton G, Fabiani M. Optical imaging of brain function. In:Parasuraman R, Rizzo M, editors. Neuroergonomics: thebrain at work. Cambridge: Oxford University Press; 2007 .p. 65—81.

8. Boas DA, Gaudette T, Strangman G, Cheng X, Marota JJA,Mandeville JB. The accuracy of near infrared spectroscopyand imaging during focal changes in cerebral hemodynamics.NeuroImage 2001;10:76—90.

9. Villringer A, Chance B. Non-invasive optical spectroscopy andimaging of human brain function. Trends Neurosci 1997;20:435—42.

10. Gratton G, Fabiani M. Shedding light on brain function: theevent-related optical signal. Trends Cogn Sci 2001;5:357—63.

11. Gratton G, Fabiani M. The event-related optical signal: a newtool for studying brain function. Int J Psychophysiol 2001;42:109—21.

12. Hintz SR, Benaron DA, Siegel AM, Zourabian A, Stevenson DK,Boas DA. Bedside functional imaging of the premature infantbrain during passive motor activation. J Prenat Med2001;29:335—43.

13. Liebert A, Wabnitz H, Steinbrink J, Moller M, MacDonald R,Rinneberg H, et al. Bed-side assessment of cerebral perfusionin stroke patients based on optical monitoring of a dye bolusby time-resolved diffuse reflectance. NeuroImage 2006;24:426—35.

14. Wilcox T, Bortfeld H, Woods R, Wruck E, Boas DA. Using near-infrared spectroscopy to assess neural activation duringobject processing in infants. J Biomed Opt 2005;10:11010.

15. Sokol DK, Markand ON, Daly EC, Luerseen TG, Malkoff MD.Near infrared spectroscopy distinguishes seizure types. Sei-zure 2000;9:323—7.

16. Watanabe E, Maki A, Kawaguchi F, Yamashita Y, Mayanagi Y.Noninvasive cerebral blood volume measurement during sei-zures using multichannel near infrared spectroscopic topo-graphy. J Biomed Opt 2000;5(3):287—90.

17. Watanabe E, Nagahori Y, Mayanagi Y. Focus diagnosis ofepilepsy using near-infrared spectroscopy. Epilepsia2002;43(Suppl. 9):50—5.

18. Watanabe E, Maki A, Kawaguchi F, Takashiro K, Yamashita Y,Koizumi K, et al. Non-invasive assessment of language dom-inance with near-infrared spectroscopic mapping. NeurosciLett 1998;256:49—52.

19. Gallagher A, Theriault M, Maclin E, Low K, Gratton G, FabianiM, et al. Language mapping using near-infrared spectroscopy(NIRS) in young epileptic patients. Epileptic Disord2007;9:241—55.

20. Gotman J, Kobayashi E, Bagshaw AP, Benar CG, Dubeau F.Combining EEG and fMRI: a multimodal tool for epilepsyresearch. J Magn Reson Imaging 2006;23(6):906—20.

21. Grova C, Daunizeau J, Lina JM, Benar CG, Benali H, Gotman J.Evaluation of EEG localization methods using realistic simu-lations of interictal spikes. NeuroImage 2006;29(3):734—53.

22. Adelson PD, Nemoto E, Scheuer M, Painter M, Morgan J, YonasH. Noninvasive continuous monitoring of cerebral oxygena-tion periictally using near-infrared spectroscopy: a prelimin-ary report. Epilepsia 1999;40(11):1484—9.

23. Haginoya K, Munakata M, Kato R, Yokoyama H, Ishizuka M,Iinuma K. Ictal cerebral haemodynamics of childhood epi-lepsy measured with near-infrared spectrophotometry. Brain2002;125:1960—71.