Abstracts

Rationale: Accurate identification of language hemisphere dominance is critical for children undergoing evaluation for epilepsy surgery. Functional MRI (fMRI) is the current gold standard, non-invasive method for presurgical brain mapping in this population. The lateralization index (LI) is frequently used in fMRI as a measure of hemisphere dominance, however, LI values are dependent on several methodological factors including the region of interest of fMRI activations, statistical threshold of fMRI activations and definitions of hemisphere dominance. Visual inspection (VI) of fMRI activation maps by experienced fMRI practitioners is an alternative method to determine language hemisphere dominance. The objective of this study is to identify the concordance between LI and VI for the determination of language hemisphere dominance in children.Methods: 52 children underwent successful fMRI for language lateralization, demonstrating reliable activation of frontal and/or temporal language regions using a minimum of 2 standardized and age adjusted language paradigms. All studies were visual rated for data quality to exclude results obscured by movement and/or poor task compliance. Once data was determined to be of acceptable quality, LI was calculated based on fMRI activations at thresholds of t = 3.0, 3.5, 4.0 and 4.5. Active voxels that were above set threshold were counted in pre-determined regions of interest for frontal and temporal language areas. All studies were also lateralized by VI by one of two experienced fMRI practitioners. Language laterality as determined by LI and VI was compared for each child. Three LI laterality definitions were used: LI ?0.2 = left hemisphere dominance; LI ? -0.2 = right hemisphere dominance and 0.2? L.I. ? 0.2 = bilateral language representation, and similarly for LI > 0.1 and LI > 0.3.Results: Using the standard LI definitions (LI ?0.2 = left hemisphere, etc.) VI and LI failed to yield the same conclusion of hemisphere dominance in 11 of the 52 cases. As multiple language paradigms were used, in some cases there was agreement on some, but not all paradigms. Varying the LI definition of laterality did not produce a significant change in the rate of discordance between LI and VI. Factors contributing to disagreement between LI and VI included the automated selection of ROI that is inherent to the calculation of the LI and a greater appreciation of bilateral foci of activation with VI.Conclusions: In the majority of cases, VI will agree with automated LI determination of hemisphere dominance for language. While LI provides a rigidly calculated measure of hemisphere dominance, it is dependent on multiple methodological assumptions. In clinical care, VI offers a more descriptive interpretation of language representation that may be of value in the assessment of individual children.