We demonstrate the feasibility of using fcMR imaging to identify epileptic foci during interictal periods in patients with medically refractory epilepsy. The fcMR imaging technique is based on calculating the intrinsic correlations of spontaneous BOLD signal across the brain to identify regions with an abnormal degree of functional connectivity. We found distinct differences in connectivity patterns in many of the patients and also found that this abnormal connectivity overlapped the epileptogenic areas identified by iEEG. This proof-of-concept suggests that fcMR imaging may be useful for identifying epileptic foci with the advantage of being noninvasive and having a high spatial resolution.
If we use the lobar localization of ictal discharges on iEEG as a reference, we can estimate the preliminary diagnostic accuracy of this method. The overall approximate sensitivity is 0.83, the specificity is 0.91, the false positive rate is 0.10, and the false negative rate is 0.15. These numbers are biased in that all of the patients tested had epileptic discharges, so a larger number of patients that do not have discharges will be required to obtain true diagnostic accuracy.
Abnormal functional connectivity has been proposed to be involved in the pathogenesis of epilepsy for decades.23
Functional connectivity as measured by fcMR imaging has been shown to be abnormal in other studies. 1,9,18,19,22,28
For example, Bettus et al.1
found a decrease in functional connectivity in affected hippocampal structures and an increase in functional connectivity between hippocampal structures in the healthy hemisphere. The methods used were different from ours in that statistical differences in the correlation coefficients for targeted regions were explored, rather than measures of local and remote coupling. Another study demonstrated disrupted functional connectivity in the language cortex in patients with temporal lobe epilepsy as compared with healthy controls.22
This change in functional connectivity might relate to plasticity of the language system due to damaged cortex, rather than a direct effect of the epileptic discharges. Moreover, Laufs et al.18,19
noted that the default network was disrupted at the group level in patients with left temporal lobe epilepsy using connectivity measures and EEG-fMR imaging. Wagner et al.28
found decreased connectivity in the hippocampus of epilepsy patients, which correlated with decreased memory performance.
Abnormal resting connectivity has also been shown in other brain disorders distinct from epilepsy, in particular Alzheimer disease, where coupling between the hippocampal formation and cortical regions is disrupted.8,10
Further exploration will be required to determine whether it is possible to separate fcMR imaging differences linked to the seizure focus from broader network alterations that may emerge from long-standing epilepsy.
Relevant to this possibility, there were bilateral increases in both local and remote coupling in 2 of our patients. This observation might indicate that the fcMR imaging is detecting abnormal functional connectivity in the entire epileptic network, as has been reported in previous fMR imaging studies (see Gotman7
for a review). When bilateral increases in connectivity are found on fcMR imaging, especially in the temporal lobe, it might be useful to compare the results with those of other neuroimaging methods such as interictal PET or ictal SPECT.
The fcMR imaging technique differs methodologically from fMR imaging with simultaneous scalp EEG, which has had a major impact recently in the neuroimaging of epilepsy.17
Functional MR imaging with simultaneous EEG inside an MR imaging scanner makes it possible to correlate the EEG with BOLD-fMR imaging.11,17
In patients with epilepsy, EEG is typically performed continuously during fMR imaging acquisition, and the images are selectively averaged based on the timing of the epileptic discharges.24
The results are sometimes ambiguous since a similar-appearing discharge may lead to an “activation” pattern (positive BOLD response) or a “deactivation” (negative BOLD response) relative to the ongoing background activity16
Abnormalities in functional connectivity patterns as measured here offer a complementary approach for identifying epileptic cortex. In this study, our hypothesis of altered BOLD connectivity reflecting epileptic discharges is supported by the spatial overlap of the fcMR imaging and iEEG, particularly for the local coupling. However, it is also interesting that remote connectivity often matches local connectivity, although not always, perhaps reflecting other network properties distant from the epileptic onset foci.
We cannot rule out the possibility that some of the connectivity abnormalities relate to structural changes due to the effects of long-standing epilepsy. Two of the patients had confirmed MR imaging–diagnosed mesial temporal sclerosis (see ). It is not clear how this might affect our results. Moreover, it is widely reported that long-standing epilepsy may result in hippocampal volume loss and cortical thinning. We have no evidence that these potential confounds exist in our results. Another limitation of the current fcMR imaging analysis is that the underlying neural activity is unknown. Simultaneous EEG would be a useful measure to determine the underlying state of the patient’s epileptic discharges at the time of the scan and would provide a more direct comparison of the two approaches in analyzing fMR imaging.