Our previous studies have demonstrated that varying fMRI paradigms may be used as probes of different aspects of language processing and that visual analysis is equivalent to quantitative region of interest analysis. ? ? Functional magnetic resonance imaging (fMRI) is increasingly used to identify eloquent cortex for pre-surgical mapping and to study the effect of pathology on the functional organization of language. However, the temporal resolution of fMRI is approximately one thousand times worse than that of electrophysiological modalities. As a result, fMRI cannot be used to explore the spectrotemporal character of neural activation. Magnetoencephalography (MEG), and particularly MEG beamformers, offer a solution by combining excellent temporal resolution with reasonable spatial resolution. Although MEG beamformers are growing in popularity, there has been relatively little research using beamformers to characterize the spectrotemporal components of language processing or to clinically map eloquent cortex. The goals of this research are twofold. The first is to develop a beamforming methodology that provides good concordance with fMRI with respect to measures of hemispheric language dominance, within anatomical regions of interest (ROIs). This will provide proof of principal. Our second goal is to then apply this methodology to characterize the time-course of linguistic cortical activation within different frequency bands. ? ? Although it is too early to give a thorough characterization of our results, some qualitative patterns are reasonably clear. First, the most robust and lateralized activation is in the theta (4-8Hz) and beta (12-20Hz) frequency bands, and somewhat less so in the high gamma (50-150Hz) band in frontal areas. Second, the most robust activation occurs in the vicinity of the semantic decision and the subjects response. Third, this method is sensitive to the temporal dynamics of the neural response within frequency bands. Fourth, right-hemisphere ROIs are often activated in the same frequency bands as the homologous left hemisphere areas. Moreover, this activation is often 500ms earlier than on the left, and almost always has the opposite sign (e.g. a decrease in power on the left and an increase in power on the right). Also, decreases in beta power, the most widely reported change in power associated with cognition, are consistently left lateralized. The pilot data suggest that MEG beamformers could be used for presurgical mapping of eloquent cortex. ? ? fMRI language tasks reliably identify language areas but memory paradigms have been problematic. We examined an auditory description decision task for hippocampal activation and compared it to the intracarotid amytal test (IAT) in 25 patients using a hippocampal mask analysis. We found that activation in hippocampus mostly reflected language dominance except where the hippocampus was sclerotic. We found poor correlation with IAT laterality indices, and only fair correlation between dominant HF activation and IAT memory capacity based on IAT. ? ? We used the paradigm of repetition priming to study the effects of epilepsy on memory. Repetition priming is characterized by improved identification of a stimulus following a previous encounter with that stimulus. This phenomenon has been linked to a decreased neuronal response commonly referred to as repetition suppression (RS). Recent evidence suggests that repetition priming may be mediated via interactions between frontal and posterior cortices. RS in posterior regions may be mediated by feedback from more anterior regions of the object processing stream. However, whether anterior temporal regions play a role in mediating RS in posterior cortices has not been evaluated. To address this question we are studying object repetition priming in patients with temporal lobe epilepsy before and after anterior temporal lobectomy. In three patients who had left temporal epileptic foci, subjects named pictures of common objects presented in a rapid event-related design. Prior to scanning subjects named 96 objects. During scanning, subjects silently named the same items (Repeated) intermixed with 96 objects not previously seen (Novel). Subjects indicated successful covert naming via a button press. The task was repeated 6-18 months following therapeutic anterior temporal lobectomy in an identical fashion but using different stimuli. ? All three patients were highly accurate at object naming both pre- and post-operatively. In addition, robust priming, as measured by button-push in the scanner, was seen pre-operatively (available for two subject), but not post-operatively (available for one subject). Most important, whereas all three patients showed robust RS bilaterally in posterior temporal and occipital cortices prior to surgery, RS was eliminated after surgery. These results suggest that RS in posterior cortices is dependent on feedback from more anterior regions of the temporal lobes. It remains to be determined whether the effect is due directly to the removal of left anterior temporal structures, or results from a disconnection of information from left inferior frontal cortex to posterior temporal regions.? ? Autosomal Dominant Partial Epilepsy with Auditory Features (ADPEAF), is caused by mutations in the LG1 gene, believed to be involved in neuronal migration. Ictal events include auditory phenomena, such as formed or unformed sounds, distortions, and sensory aphasia, and complex partial and secondary generalized tonic-clonic seizures, suggesting dominant temporal lobe pathology. We used functional magnetic resonance imaging to investigate auditory and verbal processing in patients with ADPEAF, unaffected family members, and unrelated controls. All subjects were right-handed. Each subject had 3D fast SPGR anatomical images. Monitored activation paradigms included an auditory description decision task (ASDDT), and a rising/falling musical tone discrimination task. Control conditions were reverse speech for auditory description decision, and silent rest for tone discrimination. Images were realigned, normalized, smoothed to 8 mm and analyzed with SPM2. Results were examined at p<0.05 corrected for multiple comparisons, and at p< 0.001 uncorrected. The results showed that controls showed bilateral fronto-temporal activation during tone discrimination, and predominantly left temporal and inferior frontal activation on ASDDT. Patients had decreased activation compared to normal controls on both tasks, although the effect was more marked on the semantic description decision task, where normal left posterior temporal activation was not seen, even at reduced statistical thresholds. In individual patients, activation was reduced, or displaced from control localizations. Structural MRI showed no abnormalities. Patients seizures were well-controlled on a variety of antiepileptic drugs (AEDs). Thus, our data suggest that patients with ADPEAF have disturbed functional anatomy for auditory processing. Possible AED effects cannot be excluded, but are not seen in other TLE populations. However, the LGI1 mutation may lead to cortical dysfunction in the absence of structural changes. ? ? Overall, our studies show the complex interaction of structural and functional deficts in people with epilepsy. Clinically, they highlight the need for careful individual functional mapping.
Showing the most recent 10 out of 19 publications
