European Journal of Radiology 63 (2007) 274–285 Functional MRI approach for assessing hemispheric predominance of regions activated by a phonological and a semantic task Emilie Cousin a , Carole Peyrin a , C´ edric Pichat a , Laurent Lamalle b , Jean-Franc ¸ois Le Bas b , Monica Baciu a,∗ a Laboratoire de Psychologie et Neurocognition, UMR CNRS 5105, Universit´ e Pierre Mend` es-France, BP 47, 38040 Grenoble Cedex 09, France b Unit´ e IRM, IFR1, CHU Grenoble, France Received 10 October 2006; received in revised form 20 January 2007; accepted 24 January 2007 Abstract This fMRI study performed in healthy subjects aimed at using a statistical approach in order to determine significant functional differences between hemispheres and to assess specialized regions activated during a phonological and during a semantic task. This approach (“flip” method and subsequent statistical analyses of the parameter estimates extracted from regions of interest) allows identifying: (a) hemispheric specialized regions for each language task [semantic (living categorization) and phonological (rhyme detection)] and (b) condition-specific regions with respect to paradigm conditions (task and control). Our results showed that the rhyme-specific task regions were the inferior frontal (sub-region of BA 44, 45) and left inferior parietal (BA 40, 39) lobules. Furthermore, within the inferior parietal lobule, the angular gyrus was specific to target (rhyming) items (related to successfully grapho-phonemic processing). The categorization-specific task regions were the left inferior frontal (sub-region of BA 44, 45) and superior temporal (BA 22) cortices. Furthermore, the superior temporal gyrus was related to non-target (non-living) items (correlated to task difficulty). The relatively new approach used in this study has the advantage of providing: (a) statistical significance of the hemispheric specialized regions for a given language task and (b) supplementary information in terms of paradigm condition-specificity of the activated regions. The results (standard hemispheric specialized regions for a semantic and for a phonological task) obtained in healthy subjects may constitute a basement for mapping language and assessing hemispheric predominance in epileptic patients before surgery and avoiding post-surgical impairments of language. © 2007 Elsevier Ireland Ltd. All rights reserved. Keywords: Language; Phonology; Semantics; Flip; Hemispheric dominance; fMRI 1. Introduction Mapping language areas and assessing their hemispheric pre- dominance is necessary in healthy subjects, as well as in some categories of patients. For instance, it should be performed in epileptic patients presenting focal and drug-resistant epilepsy before surgery, within the framework of the pre-surgical evalua- tion of language representation and hemispheric predominance. Within this framework, the neurosurgeons are interested not only to know which hemisphere is predominant for language but also to map specific regions underlying language opera- tions such as phonology and semantics. In these patients, the hemispheric dominance assessment of language is essential for avoiding post-surgical impairment such as aphasia. Identifying ∗ Corresponding author. Tel.: +33 476 82 78 07. E-mail address: mbaciu@upmf-grenoble.fr (M. Baciu). cerebral regions related to language operations (like semantics and phonology) is also necessary for studying cerebral reorga- nization (plasticity) of language which could occur when the epileptogenic zone is situated in the proximity of regions spe- cialized for these operations [1–4]. The neuroimaging approach became a robust method for assessing dominance and mapping language in epileptic patients [5–12]. Before mapping language in patients, the paradigms should be explored first in healthy sub- jects for identifying “standard” hemispheric specialized regions for the evaluated tasks. In this fMRI study we aimed at investigating specialized cerebral regions involved during a phonological and during a semantic paradigm. The phonological (rhyme detection) task used in this study presupposes grapho-phonemic conversion of pseudo-words in order to detect rhyming, while the semantic (living categorization) task requires not only grapho-phonemic conversion but also activation of lexical representations in order to decide if two words belong or not at the same category. We 0720-048X/$ – see front matter © 2007 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.ejrad.2007.01.030