Abstracts

While functional MRI has proven effective at mapping hemispheric dominance for language, complementing the Wada test in presurgical evaluation for intractable epilepsy, mapping of memory function has proven more difficult, as areas critical for memory are generally revealed by more subtle experimental manipulations that are mainly reliable on the group-average level. Recent evidence suggests that negative signals (deactivations) in fmri may be an important indicator of memory function. We hypothesized that task-induced decreases in the BOLD signal may reflect increased neuronal oscillatory synchronization in the theta range (4-7 Hz), which has been implicated in a variety of memory-related tasks. We used simultaneous EEG-fMRI to assess the correlation between EEG theta power modulations and hemodynamic responses evoked by a challenging memory task. Results are being compared to human intracranial recordings of the same task performed by patients undergoing presurgical monitoring. Transverse patterning is a challenging memory task shown to be selectively impaired by hippocampal damage in rodents, monkeys, and humans. We used conventional scalp EEG to assess whether the hippocampal-dependent [quot]configural[quot] condition induces greater theta power than the control [quot]elemental[quot] condition. Next, we conducted simultaneous EEG-fMRI on the same task. Locally designed hardware and software were used to reduce imaging and cardiac artifacts in the EEG data. Trial-specific theta power was used as a parametric regressor to assess correlation with hemodynamic responses. Intracranial recordings were also obtained from three surgical patients performing the same task. Induced EEG theta power was greater in the configural condition, particularly at frontal midline electrodes. Stimulus presentation induced hemodynamic responses (HR) in much of the brain. The early positive peak of the HR was generally higher in the configural condition, and the later negative undershoot was also enhanced. Trial-specific theta power correlated strongly with the magnitude of the post-stimulus BOLD undershoot in widespread portions of sensorimotor and associtional cortex. Many of these areas were seen as deactivated in block-design analysis. Intracranial recordings of task-induced theta power are also being examined, for comparison with EEG-fMRI correlation maps. Task-induced theta power, while an indicator of neural activity relevant to memory, tends to correlate negatively with the BOLD signal, at a latency associated with the post-stimulus undershoot. This suggests that BOLD undershoots may reflect meaningful neuronal activity associated with less metabolic demand and greater rhythmic synchronization. Therefore, fmri deactivations in memory tasks may offer a new window into identifying tissue critical for memory function. (Supported by NIH R01-NS38467.)