Resting-State Functional MR Imaging for Determining Language Laterality in Intractable Epilepsy

Abstract

Purpose To measure the accuracy of resting-state functional magnetic resonance (MR) imaging in determining hemispheric language dominance in patients with medically intractable focal epilepsies against the results of an intracarotid amobarbital procedure (IAP). Materials and Methods This study was approved by the institutional review board, and all subjects gave signed informed consent. Data in 23 patients with medically intractable focal epilepsy were retrospectively analyzed. All 23 patients were candidates for epilepsy surgery and underwent both IAP and resting-state functional MR imaging as part of presurgical evaluation. Language dominance was determined from functional MR imaging data by calculating a laterality index (LI) after using independent component analysis. The accuracy of this method was assessed against that of IAP by using a variety of thresholds. Sensitivity and specificity were calculated by using leave-one-out cross validation. Spatial maps of language components were qualitatively compared among each hemispheric language dominance group. Results Measurement of hemispheric language dominance with resting-state functional MR imaging was highly concordant with IAP results, with up to 96% (22 of 23) accuracy, 96% (22 of 23) sensitivity, and 96% (22 of 23) specificity. Composite language component maps in patients with typical language laterality consistently included classic language areas such as the inferior frontal gyrus, the posterior superior temporal gyrus, and the inferior parietal lobule, while those of patients with atypical language laterality also included non-classical language areas such as the superior and middle frontal gyri, the insula, and the occipital cortex. Conclusion Resting-state functional MR imaging can be used to measure language laterality in patients with medically intractable focal epilepsy. (©) RSNA, 2016 Online supplemental material is available for this article.

Figure 1:

Summary of accuracy results. Accuracy versus LI threshold at z-score thresholds of z ≥ 1, z ≥ 2, and z ≥ 3. For each patient, accuracy was computed as either 100% (eg, left functional MR imaging vs left IAP) or 0% (eg, left functional MR imaging vs right IAP). Accuracy values are averaged across subjects.

Figure 2:

LIs for each patient. Values shown are at a z-score threshold of z ≥ 2. Colors represent IAP-based laterality. Patients are shown here in the same order as in the Table. * = The patient with discordant functional MR imaging and IAP results.

Figure 3:

Composite language maps. Maps of resting-state functional MR imaging language components at z ≥ 2 are shown. Patients were grouped as having left (L), right (R), or bihemispheric (B) dominance according to functional MR imaging–based classification. Color maps = proportions and are displayed over a common anatomic template. Warmer colors = greater values. Voxels present in fewer than two patients were excluded. Images are displayed in radiologic convention. IFG = inferior frontal gyrus, IPL = inferior parietal lobule, SFG = superior frontal gyrus, STG = superior temporal gyrus.