Address correspondence and reprint requests to Graeme D. Jackson at Brain Research Institute, Neurosciences Building, Austin and Repatriation Medical Centre, Banksia Street, Heidelberg West, 3081, Victoria Australia. E-mail: BRI@brain.org.au
Summary: Objective: We performed spike triggered functional MRI (fMRI) in a 12 year old girl with Benign Epilepsy with Centro-temporal Spikes (BECTS) and left-sided spikes. Our aim was to demonstrate the cerebral origin of her interictal spikes.
Methods: EEG was recorded within the 3 Tesla MRI. Whole brain fMRI images were acquired, beginning 2–3 seconds after spikes. Baseline fMRI images were acquired when there were no spikes for 20 seconds. Image sets were compared with the Student's t-test.
Results: Ten spike and 20 baseline brain volumes were analysed. Focal activiation was seen in the inferior left sensorimotor cortex near the face area. The anterior cingulate was more active during baseline than spikes.
Conclusions: Left sided epileptiform activity in this patient with BECTS is associated with fMRI activation in the left face region of the somatosensory cortex, which would be consistent with the facial sensorimotor involvement in BECT seizures. The presence of BOLD signal change in other regions raises the possibility that the scalp recorded field of this patient with BECTs may reflect electrical change in more than one brain region.
Benign epilepsy with centro-temporal spikes (BECTS, Rolandic epilepsy) is a common childhood disorder, which usually presents with infrequent seizures, predominantly in the early evening and has an excellent prognosis for remission in adolescence (1). Typically seizures begin with numbness or paraesthesiae on one side of the mouth or cheek, followed by hemifacial twitching and difficulty speaking. At times the ipsilateral arm will have clonic movements (2), or seizures may become secondarily generalised.
The EEG features of the disorder are distinctive (3), with frequent interictal discharges seen synchronously over the central and temporal regions. They are unilateral in 60% of patients and bilateral in 40%, with a marked increase in frequency during drowsiness and sleep (4). BECTS discharges are high voltage spikes or sharp waves with maximum negativity over the centro-temporal region and lower voltage positivity over the frontal region. Source localisation has suggested that spikes are generated in the lower portion of the central sulcus (5).
Functional Magnetic Resonance Imaging (fMRI) uses Blood Oxygen Level Dependent (BOLD) contrast to reveal task-related focal cortical activiation (6,7). fMRI is also able to detect brief cerebral events, such as interictal spikes (8,9). Imaging of spikes with fMRI requires recording of the EEG while the patient is inside the MR scanner. Images are acquired when spikes are seen, and comparable images are acquired when the EEG shows no epileptiform activity.
The aim of this study was to determine the cerebral localisation of the interictal activity of a patient with BECTS using spike triggered functional MRI at 3 Tesla.
The patient was an otherwise healthy 12 year old girl seen at the first seizure clinic following a nocturnal generalised convulsion. EEG the next day showed characteristic high amplitude spike and slow wave discharges with a horizontal dipole, independently over the left and right centro-temporal regions. Background cerebral rhythms were normal, as was anatomic brain MR imaging. Ten months later, the patient was seizure free without medication and continued to have normal school performance, consistent with the electroclinical diagnosis of BECTS. A repeat EEG showed exclusively left centro-temporal spikes. The study was approved through the ethics committee of our institution and the patient gave informed consent.
EEG recorded in the MRI is exposed to more sources of artefact by the presence of the powerful magnetic field. One troublesome problem has been pulse artefact contamination. This is believed to be due to the ‘cardio-ballistic’ effect where the head pulsates subtly on the shoulders, moving electrode loops back and forth though the magnetic field, inducing currents in the EEG leads (10). Carbon fibre leads of 1.5 metres, made in-house with a resistance of 60 Ohms, were used as they are less susceptible to induced currents and may be safer, being poor conductors of heat. Eighteen non-metallic electrodes were fixed to the scalp using colloidon and conductive gel in the conventional ‘10–20’ EEG format. Two ground electrodes were placed in the midline, and two electrodes recorded ECG signal across the chest. Electrodes were twisted in pairs immediately on leaving the scalp to reduce the size of inductive loops (10), then woven in chains and taken straight out the head end of the scanner bore, thereby minimising lead movements and crossing the magnetic field lines. Head padding and restraint avoided direct pressure on scalp electrodes. The patient reported no discomfort during the study.
An EEG head-box with fibre optic coupling, transmitted the EEG signal out of the MR scanner, where it was converted back to digital for display and recording in real time. The EEG was viewed in a ‘double-banana’ bipolar montage with standard filter settings. The morphology of the patient's epileptiform discharges was characterised during a five minute recording with this equipment prior to entering the MR room (Figure 1).
MR imaging was performed on a GE (GE Medical Systems, Milwaukee, USA) Signa 3 Tesla MRI. Functional sequences used Echo-Planar Imaging (EPI BOLD) with whole brain coverage (22 slices, 4 mm thick, 1 mm gap), 128 × 128 matrix, 24 cm × 24 cm FOV, 40° flip angle, TE 40 ms, TR 3000 ms. Images were acquired in a tilted axial orientation parallel to a plane through the foramen magnum and the floor of the frontal lobe. Manual triggering, with an average delay of 2.5 seconds (range 1.8–3.0), was used to acquire a single whole brain EPI MR volume over three seconds immediately following a typical epileptiform discharge. We acquired identical single ‘baseline’ images when the EEG showed no epileptiform activity for at least 20 seconds. Spike and baseline images are acquired evenly interspersed over 60 minutes, with total MR scan time of 90 minutes.
Images were transferred to a UNIX work-station for preprocessing and analysis using Statistical Parametric Mapping (SPM99, Wellcome Department of Cognitive Neurology, http://www.fil.ion.ucl.ac.uk/spm). Preprocessing of functional images involved motion correction using rigid body transformation, spatial normalisation to Tailarach space, global intensity scaling, and smoothing with a Gaussian kernel of FWHM=6 millimeters (3 voxels). Analysis compared spike and baseline datasets for significant differences. Results, thresholded at p < 10−3 (uncorrected for multiple comparisons), were displayed as a Z-scores on the averaged EPI BOLD image to ensure activations were correctly localised. We separately examined significant increases and decreases in BOLD signal related to BECTS spikes.
Spikes were easily detectable despite the presence of some residual cardioballistic artefact on the EEG. The amplitude and morphology of the patient's epileptiform discharges was the same inside and outside the scanner (Figure 1). As the patient became drowsy, typical k-complexes, sleep spindles and later delta slowing were evident. As expected, spikes increased in frequency with drowsiness, but remained exclusively left centro-temporal throughout. fMRI volumes were triggered off 20 spikes and 24 baseline periods. Two experienced electroencephalographers later reviewed the EEG record. Due to the need for rapid decision making when acquiring ‘on-line’, some fMR images had been acquired following dubious EEG events. Only volumes triggered off spikes whose morphology and field was identical to that seen outside the MRI were included for analysis. Likewise only definite baseline acquisitions were included, confirming 10 spike and 20 baseline volumes for comparison.
Spike related BOLD signal increase was seen adjacent to the inferior left central sulcus. This is located in the face area of the left sensorimotor cortex (Figure 2). Peak activation in this area had a Z-score of 4.57, p < 0.001 (uncorrected for multiple comparisons; p = 0.06 corrected). Signal changes located around the posterior part of the brain overlay areas of magnetic field inhomogenity and were felt to be artefactual. We also looked for regions showing higher BOLD signal during baseline than spikes. This revealed spike related fMRI deactivation in the bilateral medial frontal region, adjacent to the cingulate sulcus (Figure 3), with peak Z-score of 5.18, p < 0.001 (uncorrected; p = 0.004 corrected).
We have recorded EEG, inside the 3 Telsa MRI, that permits spike triggered fMRI and demonstrated BOLD weighted signal change in association with spontaneously occurring ‘centro-temporal’ spikes. BECTS spikes in this patient were associated with fMRI activation in the left face region of the sensorimotor cortex, consistent with the facial sensorimotor involvement in BECTS seizures. In addition, deactivation was seen in the medial frontal region. It is possible that electrical field of the spikes seen in this patient with BECTS, may have a contribution from activity in more than one cortical area, rather than a single source in the Rolandic fissure.
The face/hand area of the somatosensory cortex (SSC) has previously been implicated in the pathogenesis of this condition. Seizures typically begin with parasthesiae and jerking in the mouth, face and hand, often with preserved consciousness suggesting their origin in the lower rolandic fissure (central sulcus). Imaging evidence for involvement of the SSC in BECTS spikes, has come from the subgroup of BECTS patients with ‘extreme somatosensory potentials’ (ESSP), in whom typical contralateral ‘rolandic spikes’ can be evoked by tapping or electrical stimulation of the fingers of one hand. One such subject has shown fMRI activation in the contralateral primary and secondary somatosensory and motor cortices with ESSPs (11).
BECTS spikes have a characteristic electrical field, with maximal spike negativity over the cento-temporal region and maximal positivity bifrontally. It has been felt this scalp recorded electrical field is most easily explained by a single electrical source positioned in the anterior bank of the central sulcus (5,12). Our patient with unilateral centro-temporal discharges had unilateral fMRI activations in the inferior central sulcus. However, a single electrical source is not the only possible explanation for the scalp recorded electrical field of BECTS spikes. It may be that the anterior cingulate gyrus deactivation we observe is contributing to the measured electrical field. The neurophysiologic basis of fMRI deactivation is currently unclear, but probably represents reduced neural and synaptic activity (7,13). If deactivation were to be associated with inhibitory postsynaptic potentials, effects on the scalp recorded electrical field would be possible.
The location of fMRI deactivation we observe is consistent with the supplementary motor area (14) or areas involved in attention and concentration (15). Spike triggered fMRI, by its nature, does not provide evidence on the temporal sequence of functional changes we observed, meaning we are unable to address issues of functional connectivity between these regions and their role in BECTS spike generation. Spike and baseline images were acquired interspersed throughout the experiment to minimise contamination of the results by the effects of drowsiness. However, we can not exclude the possibility that changes in alertness are influencing fMRI results. This initial study demonstrates the potential of combined EEG/fMRI as a tool to explore mechanisms of spike and seizure generation.
Acknowledgments: We acknowledge A. Simon Harvey, MD, David Reutens, MD, and Samuel Berkovic, MD, for guidance with this project and Milos Vosmansky, Jan Barchett, Louise Feiler, Jo Atkinson and Ann Godsil for advise and technical assistance with the EEG.
We would like to acknowledge the financial support of Janssen-Cilag, the National Health and Medical Research Council of Australia and the Brain Imaging Research Foundation.