Abstracts

RATIONALE: Although recent studies have indicated that functional MRI (fMRI) is a promising tool for non-invasive assessment of language lateralization, very few studies have investigated the accuracy of fMRI in locating language function in sufficient detail 1,2. METHODS: Ten patients underwent electrocortical stimulation mapping (ESM) during epilepsy surgery. Prior to their operation patients participated in four different, visually presented fMRI language tasks. Scans were obtained using a navigated PRESTO sequence that previously demonstrated a close match between ESM language foci and fMRI signal change.2 Craniotomy exposed inferior frontal and temporoparietal areas. During ESM patients performed a picture naming task. The entire ESM procedure was recorded with a video camera mounted to the operating lamp, approximately 50 cm above the patient's head. Afterwards, ESM positions were digitized and a computer program counted the number of fMRI tasks that yielded active voxels within a radius of 6 mm from ESM sites. RESULTS & CONCLUSION: A total of 400 cortical stimulations was performed at 232 different sites. Combining activity maps from the four fMRI tasks succesfully identified 36 out of 37 ESM-detected language areas with high accuracy (< 6 mm). Individual fMRI tasks did not meet this high standard, indicating that combining statistical maps from several tasks significantly improved detection rate. Negative ESM findings, i.e. no language errors during stimulation, were matched by negative fMRI results in only 60 % of foci. New algorithms to improve this match are investigated. Our findings indicate that fMRI detection of language-critical regions, although it may not yet be adequate to replace ESM, benefits from a multiple task design. REFERENCES: 1. Fitzgerald DB et al. ANJR 1997; 18: 1529-1539 2. Rutten GJM et al. Annals of Neurology 1999; 46: 405-408