Enhancing fMRI language lateralization indices by homotopic mapping

Abstract The non-invasive nature of fMRI would make it an excellent tool for determining language lateralization, potentially replacing the WADA test that is now considered the gold standard. However, while the result of the WADA test has 3 straightforward categories, i.e. left, right, or bilateral language lateralization, fMRI lateralization indices (LIs) are on a continuous scale between −1 (completely right lateralized) and 1 (completely left lateralized), requiring arbitrary thresholding to come to a useful clinical measure. We hypothesize that the absence of clear categories in fMRI LIs is in part linked to imperfect fMRI task control and/or homotopically mapped language activity in the non-dominant hemisphere. In this study we attempt to account for these sources of brain activity by making a detailed homotopic comparison between voxels in the language areas and their homologues, which is now possible using the Cgrid toolbox. 36 epilepsy were included who underwent presurgical fMRI including a picture naming and verb generation task, in addition to the WADA test, functional Transcranial Doppler sonography (fTCD), or Electrocortical stimulation (ECS). LIs were calculated using a traditional approach, and a novel approach based on the paired comparison between language voxels and their homologues. We observed that lateralization indices came in closer alignment with those established by the WADA test, ECS, of fTCD. Activity maps indicated that the accounting for imperfect fMRI task control was at least part of the reason for the improvement when using the new approach. This Cgrid based approach could help fMRI replacing the more invasive WADA test in the future.