Abstracts

Rationale: Coherence is a measure of synchronization between brain regions. Synchronized activity within a neuronal network is determined by the strength of network connections. How well two or more brain regions are connected can be determined by measuring the coherence between these regions. This study assessed the use of MEG imaging of brain coherence in presurgical patients with refractory epilepsy, to provide localizing data concordant with the epileptogenic zone.Methods: A retrospective analysis of MEG data acquired in the presurgical evaluation of patients with refractory partial epilepsy was completed. Only patients who were seizure-free with at least 2 years of postsurgical follow-up were included. MEG data were analyzed using independent component analysis (ICA) to extract independent sources of brain activity, these sources were then imaged with MR-FOCUSS (current density). For active sources, a sequence of FFT spectra were calculated (0.5 second windows, 25% overlap) and a coherence spectral matrix between sources was calculated for each frequency below 50 Hz. Brain maps showing high coherence were compared to the region of surgical resection on post-operative MRI studies, and also to images from normal controls. MEG coherence images were generated from 10-15 minutes of spontaneous brain activity in no specific state (awake, drowsy and/or sleep). Results: Twenty-six epilepsy patients (8 left temporal lobe epilepsy, 16 right temporal lobe epilepsy, 2 right frontal lobe epilepsy) were identified who were seizure-free with postsurgical follow-up ranging from 24-78 months. Increased coherence (coregistered on presurgical MRI) was localized to the area of resection in 19/24 temporal lobe cases. Two temporal lobe cases showed bitemporal coherence changes but prominence could not be determined. Two other cases showed a pattern of bitemporal coherence, but coherence appeared more prominent contralateral to the site of resection. In one other temporal lobe case, the increased coherence lateralized correctly to the right hemisphere, but did not appear maximal in the temporal region. In 3 of 4 cases which could not be lateralized or incorrectly lateralized, coherence imaging suggested left greater than right temporal involvement, similar to a pattern seen in some normal controls. In the two frontal lobe cases increased coherence did not localize to the resected site. Conclusions: Imaging high coherence between brain regions may be used to identify zones of epileptic activity characterized by abnormally synchronized neuronal activity within a distributed network. MEG and EEG are sensitive to small changes in synchrony (coherence) within neuronal populations, but these changes do not necessarily require increased metabolism, and would be undetectable in fMRI and PET recordings. (This research was supported by NIH/NINDS Grant R01NS 30914.)