A practical clinical method to quantify language lateralization in fMRI using whole-brain analysis

Surgery is often the only effective treatment for intractable epilepsy, but its benefits must be balanced by potential disruption of eloquent cortical functions. Wada test is the standard technique to lateralize language before surgery; however, it is invasive and associated with complications. fMRI provides an attractive noninvasive alternative, which has been previously shown to correlate with Wada results. However this correlation is imperfect since standard fMRI laterality indices are dependent on a particular arbitrary statistical threshold used in the data processing. We report a novel automated, threshold-independent fMRI methodology to assess language lateralization, which we hypothesize provides a robust and unbiased pre-operative assessment. This hemispheric histogram analysis method can accurately interrogate language lateralization, as validated against the Wada test. Fifty-nine subjects with intractable epilepsy received preoperative evaluation for language lateralization using fMRI. fMRI data then were analyzed using a novel automated threshold-independent method for determining language lateralization. The methodology generated a lateralization score based on hemispheric activation of language areas and a quality index based on multiple factors, including patient motion and signal-to-noise characteristics. Lateralization scores were compared to Wada test results (51 patients), direct cortical stimulation (3 patients), and subdural grid stimulation (5 patients). Data sets were used to generate a probability score for language lateralization for each subject. The lateralization scores correlated well with the objective measures of language lateralization (r(2)=0.46). Cumulative historical data were utilized to prospectively determine probabilities of language lateralization for individual patients. In conclusion, hemispheric language lateralization can be accurately determined using a novel objective and automated methodology that calculates language lateralization in a threshold-independent manner and can be used to determine the probability of language dominance in individual patients.