Abstracts

Rationale: Presurgical language assessment can help minimize damage to eloquent cortex during resective epilepsy surgery. Two methods for language mapping are functional MRI and direct cortical stimulation (DCS) with implanted subdural electrodes. FMRI is noninvasive and cost effective and is supplanting the intracarotid amytal test for language lateralization in many institutions, but not DCS. We examined the relationship between fMRI and DCS in language mapping to evaluate the utility of language fMRI tasks as tools for helping plan electrode implantation, and potentially provide independent language mapping data Methods: We studied 18 patients referred to the Clinical Epilepsy Section, National Institute of Neurological Disease and Stroke, National Institutes of Health for presurgical evaluation of drug-resistant epilepsy. Patients completed 4 language tasks during preoperative fMRI. After subdural electrode implantation, we used DCS to localize language areas. Electrodes showing language errors on stimulation were considered to overlie a language critical region of cortex, and were defined as language positive. Subdural electrode locations were determined by extracting coordinates from postoperative CT, registration to preoperative MRI, and projection to each patient's cortical envelope. Each electrode pair was transformed into an ROI and overlaid on patients' fMRI activation maps. For each stimulation site, an algorithm determined whether electrode pair ROIs fell within significant fMRI activity clusters for each language task. This test was corrected for multiple comparisons based on Monte Carlo simulation (Ward, 2000), which established correction thresholds for 3 p values (in order from liberal to strict: p < 0.05, 0.01, 0.001). We included ROIs of 3 sizes (4, 6, and 8 mm) to account for uncertainty in effective cortical stimulation and propagation radii. 403 electrodes were considered in total, with 49 being identified as language positive by DCS. Results: Sensitivity and specificity depended heavily on electrode ROI radii and statistical thresholding. As expected, sensitivity increased with increasing radius size and decreased with stricter thresholding. Figure 1 shows sensitivities and specificities for each parameter combination. For patients with at least one language positive stimulation site, the auditory description task evaluated with an 8 mm ROI radius and strict threshold (p=0.001, uncorrected) offered the best combination of sensitivity (62%) and specificity (87%). For patients with no language positive stimulation sites, fMRI was a highly specific method of ruling out critical language areas. Conclusions: Language fMRI is an effective tool for determining language lateralization prior to electrode implantation, and is especially useful for excluding unexpected critical language areas. It can help guide subdural electrode implantation and narrow the search for eloquent cortical areas via DCS. Funding: NINDS NIH Division of Intramural Research