Abstracts

Rationale: The localization of interictal activity in epilepsy using functional MRI methods that do not require simultaneous EEG may provide information regarding activity that is not detected with scalp EEG. In this work a data-driven fMRI image analysis method designed to detect transient spiking of a signal, 2dTCA [1-2], was used to localize interictal activity in a group of temporal lobe epilepsy patients and investigate brain networks from one activated region. Methods: Five patients with left mesial temporal sclerosis or hippocampal structural abnormality who were part of a larger investigation [2] were reanalyzed for this study. All underwent left selective hippocampectomy and became free of disabling epileptic seizures. Subjects were imaged on a 3.0T MRI scanner including structural and fMRI scans at rest with eyes closed (64x64, FOV=240 mm, TE/TR=35/2000 ms, 200 volumes, 6:40 scan time). The 2dTCA analysis was performed on the group to detect transient fMRI signal spiking. Different temporal spiking profiles were separated by the algorithm. All temporal spiking profiles were included as regressors with the global time course and motion parameters in a general linear model (GLM) in SPM5 software to localize the activation corresponding to each timing profile. The average time course in one region of activation in the left hippocampus was then used as a seed for connectivity analyses implemented with the GLM in SPM5. Results: The 2dTCA analysis yielded four unique spike timing profiles of transient activation resulting in four activation maps: (1) bilateral temporal regions consistent with auditory activation, (2) left superior parietal and dorsolateral prefrontal regions, (3) regions of the default-mode network [3], and (4) left hippocampus (Fig 1a). The first three have been detected as normal resting-state networks using ICA [4] and any possible relationship of their transient activity to the interictal activity is unknown. The activation in the left hippocampus is consistent with other measures of interictal epileptogenicity in these patients. The connectivity map revealed strong correlations between the left and right hippocampus in this group (Fig 1b). It should be noted that neither negative activation nor correlation in the connectivity analyses was detected in the default-mode network as reported using fMRI-EEG [5]. Conclusions: This work shows that interictal activity can be localized using fMRI with 2dTCA with possibly more specificity (less components) than ICA [6]. This method may provide unique information from that found using simultaneous fMRI-EEG. Finally, these activated regions may then be used to determine possible epileptogenic networks. This work was supported by the Epilepsy Foundation, NIH EB00461 and NIH NS055822. 1. Morgan VL, et al., Hum Brain Mapp 2008;29:57. 2. Morgan VL, et al., Epilepsy Research 2007;76:22. 3. Raichle ME, et al. PNAS 2001;98:676. 4. Damoiseaux JS, et al., PNAS 2006;103:13848. 5. Laufs H, et al. Hum Brain Mapp 2007 ;28 :1023. 6. Rodinov R, et al. NeuroImage 2007;38:488.