Abstracts

Simultaneous measurement of electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) shows great promise as a clinical tool and as a method to study the origin of interictal spiking^1,2. However, a serious impediment to its routine use is the relatively low percentage of patients who show fMRI activations despite having interictal discharges whilst in the scanner. The purpose of the current work was to attempt to increase the number of patients with significant activations by analysing the fMRI data with a range of modelled haemodynamic response functions (HRFs).
fMRI images were acquired in one of two 1.5T MR scanners (Vision and Sonata, Siemens, Germany) using echo-planar imaging. Twenty-one channels of EEG were recorded simultaneously using the EMR amplifier (Schwarzer, Germany). Statistical processing of the fMRI images produced maps of the t statistic, allowing positive and negative activations to be detected³. Thirty-four image sets were selected from a total scanned population of 42 patients, on the basis of a clinical diagnosis of focal epilepsy and more than 5 spikes during the fMRI session. The initial analysis was performed using the measured HRF to brief auditory stimuli which peaks at 5.4 seconds⁴. The data were reanalysed with four alternative HRFs modelled as single gamma functions peaking at 3, 5, 7 and 9 seconds. The source of epileptic activity was determined from EEG and clinical data.
Twelve data sets (35%) showed activations that were in agreement with the known source of the spikes and which had a higher t value when using one of the single gamma functions than when using the original HRF. Two sets demonstrated areas of activation that were also consistent with the focus but which had a higher t value when using the original HRF. In 3 sets the original and new analyses were very similar, and in 4 sets areas of activation which were not consistent with the known source were found when using one of the single gamma function HRFs. In the remaining 13 (38%) sets, no activation was seen.
Thirteen data sets showed areas of negative activation. All but two, both from the same subject, also had some areas of significant positive activation. In 10 of the 13, the most significant negative cluster had a maximum t value when using an HRF that peaked later than the original. In comparison, the positive activations had maximum t values that were evenly spread across the five HRFs.
The results suggest that an exploratory approach based upon using several HRFs with different latencies can provide new information concerning the source of interictal epileptic spikes. However, misrepresentation of the form of the HRF is not the major reason for the high percentage of subjects who do not show fMRI activation following interictal discharges.
¹Bénar CG et al 2002 NeuroImage 17 1182
²Lemieux L et al 2001 NeuroImage 14 780
³Worsley KJ et al 2002 NeuroImage 15 1
⁴Glover GH 1999 NeuroImage 9 416