Language function shows comparable cortical patterns by functional MRI and repetitive nTMS in healthy volunteers

In preoperative planning, fMRI and repetitive navigated transcranial magnetic stimulation (rTMS) repeatedly revealed differences in the detected language sites, which can be attributed to tumor-induced oxygenation changes impairing the accuracy of fMRI. We therefore compared the accordance of those techniques in healthy subjects using exactly the same tasks in both investigations. 19 healthy right-handed subjects performed object naming, pseudoword reading, verb generation, and action naming during fMRI at 3T and rTMS. For rTMS language mapping, we stimulated 46 cortical spots over the left hemisphere; each site was stimulated for three times. Language positive points during rTMS for one, two, or three errors out the three stimulations per spot (1/3, 2/3, 3/3) were exported via DICOM, and compared to the positive fMRI clusters. As a result of this comparison, the best correlation was observed between 3/3 errors and fMRI for pseudoword reading and verb generation with t-values of p(u)<0.001, uncorrected for multiple comparisons, on average across the whole rTMS-spot map. We found a close spatial agreement between several rTMS-spots (2/3 and 3/3 errors) and fMRI clusters accentuated in the frontal lobe, followed by the parietal lobe and less in the temporal lobe. Compared to the fMRI clusters, there was a higher congruence for 2/3 and 3/3 errors than for 1/3 errors. Overall, results of language mapping in healthy subjects by fMRI and rTMS correspond well yet depending on the used language task.