Neuropediatrics 2015; 46 - FV03-02
DOI: 10.1055/s-0035-1550662

A Comparison of Language Lateralization as Assessed using Functional MRI using Scanner-Generated versus Externally Calculated Statistical Parameter Maps

M. Boelen 1, A. Zsoter 2, K. Lidzba 1, M. Staudt 2, M. Wilke 1
  • 1Klink für Kinder- und Jugendmedizin, Abteilung III (Neuropädiatrie), Tübingen, Germany
  • 2Klinik für Neuropädiatrie und Neurologische Rehabilitation, Epilepsiezentrum für Kinder und Jugendliche, Vogtareuth, Germany

Aims: In the recent years, the preoperative assessment of language functions using functional MRI (fMRI) has become increasingly important in clinical practice, also for pediatric patients. Here, interpretation of results usually relies on scanner-generated statistical parameter maps. However, much more extensive processing approaches are used in the research arena. This study was aimed to directly compare these different approaches.

Methods: For this project, we used fMRI data sets collected in a pediatric epilepsy center from January 2010 to September 2014. Demographic information of the patients was collected, and data sets from the vowel identification task were processed using SPM12 software (FIL, University College, London) running in MATLAB. After an extensive preprocessing and final spatial smoothing (FWHM = 4 mm), statistical parameter maps were generated using the general linear model. These, as well as the automatically generated statistical maps from the MRI scanner, were then assessed regarding their lateralization in the frontal lobe using a robust, threshold-free lateralization index (LI).

Results: There were 173 series of 84 patients (median age, 13.49 [MAD, 2.81] years; sex, M 45 and W 39; handedness, L 22 and R 56, ambidextrous 4, and unknown 2). The results of both approaches were very similar, with LI = 0.35 (0.35) versus LI = 0.30 (0.35). The difference was not significant (p = 0.3; Mann–Whitney U test). However, several individual data sets showed substantial differences when comparing the two approaches, which would also lead to a change in the interpretation of their hemispheric dominance.

Conclusion: The comparison of scanner-generated versus externally calculated statistical maps yields only minor differences on the group level. However, there are several individual cases with substantial discrepancies. The analysis of the underlying factors (data quality, subject movement, etc.) is currently ongoing.

Keywords: functional MRI, language lateralization, hemispheric dominance.