Concordance of verbal memory and language fMRI lateralization in people with epilepsy

Background and purpose: This work investigates verbal memory functional MRI (fMRI) versus language fMRI in terms of lateralization, and assesses the validity of performing word recognition during the functional scan.

Methods: Thirty patients with a diagnosis of epilepsy underwent verbal memory, visuospatial memory, and language fMRI. We used word encoding, word recognition, image encoding, and image recognition memory tasks, and semantic description, reading comprehension, and listening comprehension language tasks. We used three common lateralization metrics: network spatial distribution, maximum statistical value, and laterality index (LI).

Results: Lateralization of signal spatial distribution resulted in poor similarity between verbal memory and language fMRI tasks. Signal maximum lateralization showed significant (>.8) but not perfect (1) similarity. Word encoding LI showed significant correlation only with listening comprehension LI (p = .016). Word recognition LI was significantly correlated with expressive language semantic description LI (p = .024) and receptive language reading and listening comprehension LIs (p = .015 and p = .019, respectively). There was no correlation between LIs of the visuospatial tasks and LIs of the language tasks.

Conclusions: Our results support the association between language and verbal memory lateralization, optimally determined by LI quantification, and the introduction of quantitative means for language fMRI interpretation in clinical settings where verbal memory lateralization is imperative.