Increasing the validity of functional MRI for presurgical language lateralization: proposal for an algorithm

Introduction: In the presurgical evaluation of epilepsy patients, functional MRI (fMRI) increasingly replaces the intracarotid amobarbital procedure (IAP) for language lateralization. However, recent studies reveal relevant discrepancy between both methods on the single patient level. This questions the uncritical application of fMRI as the only tool for language lateralization. Conditions have to be known, under which fMRI may lateralize language incorrectly. Materials and Methods: 238 patients received fMRI, 65 of them also IAP. By applying the criteria (1) absence of fMRI-interfering lesions, (2) sufficient patient's compliance inside the scanner, and (3) application of lateralization indices high enough to identify unilateral language dominance in accordance with IAP we identified patients feasible for isolated fMRI-based language lateralization. Application of an algorithm comprising these criteria to 238 patients allows to estimate the value of fMRI for presurgical language lateralization. Results: In our series, 22.6% of patients had lesions potentially interfering with fMRI, 4.7% showed compliance questioning sufficient attention to the given task. 60.3% of the patients failed to show the protocol specific lateralization index that guarantees for complete concordance between fMRI and IAP. According the proposed algorithm, fMRI was safe enough to be the only tool for language lateralization in only 32% of patients. Conclusions: This algorithm represent a conservative approach to clinical application of fMRI. However, when fMRI is used for presurgical language lateralization on the single patient level, it's specificity should be maximized even on cost of it's sensitivity. In patients with doubtful fMRI, IAP should be performed.